Beruflich Dokumente
Kultur Dokumente
o o
o
11 Years of Progress
Project Managed by the Jet Propulsion Laboratory for the U.S. Department of Energy
75 10
MODULE PRICE
~
/
--- -- ---
«
L!J MODULE LIFE
Cl... 10 #
f- >- (f)
:;;: u 0:
Z «
~
L!J
w
<f>- U >-
L!J
U
5 l±
w
0:
Cl...
5
1975 1986 1975 1986
Flat or non-concentrating module prices have dropped Typical module lifetimes were less than 1 year but
as module efficiencies have increased. Prices are in are now estimated to be greater than 10 years.
1985 dollars for large quantities of commercial products. (Ten-year warranties are now available.)
Union Carbide Corporation (UCC) funded the now The automated machine interconnects solar cells
operational silicon refinement production plant with and places them for module assembly. The second-
1200 MTlyear capacity. DOEIFSA-sponsored efforts generation machine made by Kulicke and Sotta was
were prominent in the VCC process research cost shared by Westinghouse Corporation and DOEIFSA
and development.
I I II
I
II: ti
Prepared for
U.S. Department of Energy
Through an Agreement with Project Managed
National Aeronautics and Space Administration by the
by Jet Propulsion Laboratory
Jet Propulsion Laboratory for the
California Institute of Technology U.S. Department of Energy's
Pasadena, California National Photovoltaics Program
FSA Project: 10 Years of Progress, JPL Document 400-279, 5101-279, October 1985.
Summary of FSA Project Documentation: Abstracts of Published Documents, 1975 to 1986, JPL Publication 82-79
(Revision 1), 5101-221, DOE/JPL-1 012-76, September 1986.
Upon request, this FSA Final Report (JPL Publication 86-31) and the two supplemental reports (JPL Document
400-279 and JPL Publication 82-79) are individually available in print from:
ii
Abstract
The Flat-Plate Solar Array (FSA) Project, funded by the U.S. Government and managed by the Jet Propulsion
Laboratory, was formed in 1975 to develop the module/array technology needed to attain widespread terrestrial
use of photovoltaics by 1985. To accomplish this, the FSA Project established and managed an Industry, Univer-
sity, and Federal Government Team to perform the needed research and development.
The objective of the Encapsulation Task was to develop, demonstrate, and qualify photovoltaic (PV) module
encapsulation systems that would provide 20-year (later increased to 30-year) life expectancies in terrestrial en-
vironments, and which would be compatible with the cost and performance goals of the FSA Project. The scope
of the Encapsulation Task included the identification, development, and evaluation of material systems and con-
figurations required to support and protect the optically and electrically active solar cell circuit components in
the PV module operating environment. Encapsulation material technologies summarized in this report include the
development of low-cost ultraviolet protection techniques, stable low-cost pottants, soiling resistant coatings,
electrical isolation criteria, processes for optimum interface bonding, and analytical and experimental tools for
evaluating the long-term durability and structural adequacy of encapsulated modules. Field testing, accelerated stress
testing, and design studies have demonstrated that encapsulation materials, processes, and configurations are
available that will meet the FSA cost and performance goals. Thirty-year module life expectancies are anticipated
based on accelerated stress testing results and on extrapolation of real-time field exposures in excess of 9 years.
iii
Foreword
Throughout U.S. history, the Nation's main source of energy has changed from wood to coal to petroleum. It is
inevitable that changes will continue as fossil fuels are depleted. Within a lifetime, it is expected that most U.S. energy
will come from a variety of sources, including renewable energy sources, instead of from a single type of fuel. More
than 30 % of the energy consumed in the United States is used for the generation of electricity. The consumption of
electricity is increasing at a faster rate than the use of other energy forms and this trend is expected to continue.
Photovoltaics, a promising way to generate electricity, is expected to provide significant amounts of power in years to
come. It uses solar cells to generate electricity directly from sunlight, cleanly and reliably, without moving parts.
Photovoltaic (PV) power s"ystems are simple, flexible, modular, and adaptable to many different applications in an
almost infinite number of sizes and in diverse environments. Although photovoltaics is a proven technology that is
cost-effective for hundreds of small applications, it is not yet cost-effective for large-scale utility use in the United
States. For widespread economical use, the cost of generating power with photovoltaics must continue to be
decreased by reducing the initial PV system cost, by increasing efficiency (reduction of land requirements), and by
increasing the operational lifetime of the PV systems.
In the early 1970s, the pressures of the increasing demand for electrical power, combined with the uncertainty of
fuel sources and ever-increasing prices for petroleum, led the U.S. Government to initiate a terrestrial PV research and
development (R&D) project. The objective was to reduce the cost of manufacturing solar cells and modules. This
effort, assigned to the Jet Propulsion Laboratory, evolved from more than a decade-and-a-half of spacecraft PV power-
system experience and from recommendations of a conference on Solar Photovoltaic Energy held in 1973 at Cherry
Hill, New Jersey.
This Project, originally called the Low-Cost Solar Array Project, but later known as the Flat-Plate Solar Array (FSA)
Project, was based upon crystalline-silicon technology as developed for the space program. During the 1960s and
1970s, it had been demonstrated that photovoltaics was a dependable electrical power source for spacecraft. In this
time interval, solar-cell quality and performance improved while the costs decreased. However, in 1975 the costs were
still much too high for widespread use on Earth. It was necessary to reduce the manufacturing costs of solar cells by a
factor of approximately 100 if they were to be a practical, widely used terrestrial power source.
The FSA Project was initiated to meet specific cost, efficiency, production capacity, and lifetime goals by R&D in all
phases of flat-plate module (non-concentrating) technology, from solar-cell silicon material purification through verifica-
tion of module reliability and performance.
The FSA Project was phased out at the end of September 1986.
iv
Acknowledgments
Authorities in the field were used throughout this program as consultants to provide critiques, participate
in critical technology reviews, and address specific technical problems. Otto Vogi of the University of
Massachusetts, and James Guillet of the University of Toronto were consultants in photodegradation and its
control. Jack Koenig of Case Western Reserve University, James Boerio of the University of Cincinnati,
Edwin Plueddemann of Dow Corning in Midland, Michigan, and Wolfgang Knauss of the California Institute
of Technology consulted in the area of interfacial bonding.
The Jet Propulsion Laboratory staff members were involved in laboratory research, materials technology
development, process and design analyses, and the technical management of contracts. The following
engineers contributed to the encapsulation technology program: Clifford Coulbert (Task Manager), Andre
Amy, Ed Cuddihy, Ami Gupta, John Repar, Ranty Liang, Dan Coulter, Jovan Moacanin, Peter Frickland, Keri
ada, William Carroll (deceased), Frank Bouquet, Robert Holtze, Hugh Maxwell, Mohammad Sarbolouki, John
Garba, and Harold Marsh.
This document reports on work done under NASA Task RE-152, Amendment 419, DOE/NASA IAA No.
DE-A 101-85CE89008.
v
FSA Project Summary
The Flat-Plate Solar Array (FSA) Project, a Government-sponsored photovoltaic (PV) project, was initiated in
January 1975 with the intent to stimulate the development of PV systems for economically competitive, large-
scale terrestrial use. The Project's goal was to develop, by 1985, the technology needed to produce PV modules
with 10% energy conversion efficiency, a 20-year lifetime, and a selling price of $O.50/W p (in 1975 dollars). The
key achievement needed was cost reduction in the manufacture of solar cells and modules.
As manager, the Jet Propulsion Laboratory organized the Project to meet the stated goals through research and
development (R&D) in all phases of flat-plate module technology, ranging from silicon-material refinement through
verification of module reliability and performance. The Project sponsored parallel technology efforts with periodic pro-
gress reviews. Module manufacturing cost analyses were developed that permitted cost-goal allocations to be made
for each technology. Economic analyses, performed periodically, permitted assessment of each technical option's
potential for meeting the Project goal and of the Project's progress toward the National goal. Only the most promising
options were continued. Most funds were used to sponsor R&D in private organizations and universities, and led to
an effective Federal Government-University-Industry Team that cooperated to achieve rapid advancement in PV
technology.
Excellent technical progress led to a growing participation by the private sector. By 1981, effective energy conser-
vation, a leveling of energy prices, and decreased Govemment emphasis had altered the economic perspective for
photovoltaics. The U.S. Department of Energy's (DOE's) National Photovoltaics Program was redirected to longer-
range research efforts that the private sector avoided because of higher risk and longer payoff time. Thus, FSA con-
centrated its efforts on overcoming specific critical technological barriers to high efficiency, long life, reliability, and
low-cost manUfacturing.
To be competitive for use in utility central-station generation plants in the 1990s, it is estimated that the price of
PV-generated power will need to be $O.17/kWh (1985 dollars). This price is the basis for a DOE Five-Year Photo-
voltaics Research Plan involving both increased cell efficiency and module lifetime. Area-related costs for PV utility
plants are significant enough that flat-plate module efficiencies must be raised to between 13 and 17%, and module
life extended to 30 years. Crystalline silicon, research solar cells (non-concentrating) have been fabricated with more
than 20% efficiency. A full-size experimental 15% efficient module also has been fabricated. It is calculated that a
multimegawatt PV power plant using large-volume production modules that incorporate the latest crystalline silicon
technology could produce power for about $0.27/kWh (1985 dollars). It is believed that $O.17/kWh (1985 dollars) is
achievable, but only with a renewed and dedicated effort.
Government-sponsored efforts, plus private investments, have resulted in a small, but growing terrestrial PV in-
dustry with economically competitive products for stand-alone PV power systems. A few megawatt-sized, utility-
connected, PV installations, made possible by Government sponsorship and tax incentives, have demonstrated the
technical feasibility and excellent reliability of large, multimegawatt PV power-generation plants using crystalline sili-
con solar cells.
e Established basic technologies for all aspects of the manufacture of nonconcentrating, crystalline-silicon PV
modules and arrays for terrestrial use. Module durability also has been evaluated. These resulted in:
• Reducing PV module prices by a factor of 15 from $75/Wp (1985 dollars) to $5/Wp (1985 dollars).
• Increasing module efficiencies from 5 to 6% in 1975 to more than 15% in 1985.
e Stimulating industry to establish 10-year warranties on production modules. There were no warranties in 1975.
.. Establishing a new, low-cost high-purity silicon feedstock-material refinement process.
e Establishing knowledge and capabilities for PV module/array engineering/design and evaluation.
• Establishing long-life PV module encapsulation systems.
• Devising manufacturing and life-cycle cost economic analyses.
.. Transferred technologies to the private sector by interactive activities in research, development, and field
demonstrations. These included 256 R&D contracts, comprehensive module development and evaluation efforts,
26 Project Integration Meetings, 10 research forums, presentations at hundreds of technical meetings, and ad-
visory efforts to industry on specific technical problems.
• Stimulated the establishment of a viable commercial PV industry in the United States.
vi
Module Encapsulation Summary
The Flat-Plate Solar Array (FSA) Encapsulation Task had two major objectives during its 11-year activity
span. The first was the identification, development, evaluation, and qualification of photovoltaic (PV) module
encapsulation systems and fabrication processes that would meet the FSA Project cost, performance, and
durability goals. The second was the achievement of an understanding and quantification of encapsulation
degradation mechanisms and the development of techniques for eliminating or controlling degradation in
order to attain a 30-year module lifetime.
The FSA Project, in its analysis of requirements for the achievement of cost-effective PV module
designs, developed specific performance goals and cost allocations for each of the module technologies.
The following goals were defined for the encapsulation task:
• Cost of encapsulation materials, including substrate and/or superstrate, not exceeding $1 .30!ft 2
(1980 dollars).
• Initial optical transmission of 90 %, and loss of less than 5 % after 30 years of use.
Following initial definition of the performance and cost requirements and identification of expected
environmental hazards, an extensive survey of many possible candidate material systems and processes
was conducted. It was concluded that new or modified low-cost materials (especially polymers) and
improved processing techniques would be required to meet the Project goals. The available polymer
materials considered weatherable (such as silicones, fluorocarbons, and acrylics) were either too expensive
or not available in a suitable form for pottant or outer cover application.
Early module encapsulation material systems experienced several typical failure modes that required
both new materials and revised design configurations. The more significant problems included: delamination,
water penetration, surface soiling, substrate deterioration, short circuits, cracked cells, and interconnect
fatigue failures. This report reviews how these problems were resolved.
The resolution of these problems required research and development (R&D) in the areas of polymer
synthesis, photodegradation mechanisms, ultraviolet stabilizer synthesis, accelerated testing techniques,
primer and adhesive technology, unique processing concepts, and anti-soiling surface treatments. Analytical
modeling studies (including the development of new computer programs) were undertaken to understand
and predict thermal, structural, optical, electrical, and photodegradation effects as a function of materials,
module configuration, and time. This report reviews the development of these models and how they were
used in the design of PV module encapsulation systems.
Working relationships and technical exchange interfaces were established with the PV-module and
material-supply industries as well as with the academic and Government R&D organizations. A number of
subcontracts were established to support these efforts. Cooperative exchanges of requirements and capa-
bilities information that occurred with major material suppliers aided in the development by industry of new
material products and in the transfer of technology to module manufacturers.
.. Identification and evaluation of suitable candidate materials and manufacturing processes for each
element of a complete module encapsulation system, including front-cover su perstrates pottants,
adhesives, back covers, substrates, and edge seals with projected costs less than $1.30/ft 2
(1980 dollars).
vii
• Development of anti-soiling treatments for module covers that have reduced light transmission losses
to less than 2 % from more than 8 % without treatment.
• Development and test verification of module design procedures and design tools for PV modules that
meet the Project cost and performance goals.
• Development and verification of accelerated testing equipment and techniques for encapsulation
materials subjected to photodegradation and long-term environmental stresses.
• Establishment of technical interfaces with the major U.S. materials suppliers (e.g., Dow Corning,
Du Pont, 3M, General Electric, and Gulf Oil), resulting in commercialization of the more promising
encapsulant candidates such as EVA.
• Effective transfer of encapsulation technology to industry and the PV module manufacturers' response
in adopting it for their new product lines.
• Development of encapsulation systems with electrical breakdown voltages greater than 5000 V DC.
• Verification by accelerated tests of the probability of 30-year PV module lifetimes. Several PV module
manufacturers are now granting 10-year warranties.
viii
Contents
I. INTRODUCTION.
A. TASK OBJECTIVES.
B. TASK GOALS
C. BACKGROUND.
A. BACKGROUND. 3
1. Superstrates/Substrates . 3
2. Pottants . 4
4. Back Covers · 10
6. Porous Spacer. · 10
C. HYDROTHERMAL AGING . . 13
D. CORROSION PREVENTION. . 14
B. LOW-SOILING COATINGS .. 19
A. THERMAL-OPTICAL MODELING. 25
Ix
B. STRUCTURAL MODELING 34
1. Glass Superstrate Design. 34
2. Mild Steel Substrate Design .. 36
3. Wood-Hardboard Substrate Designs 36
4. Master Curves for Structural Analysis. 37
5. Aluminum Panel .. 39
6. Wood Panel. 39
C. ELECTRICAL INSULATION (SAFETY) . 41
1. Voltage-Breakdown Data. 42
2. Electrical Insulation Aging 43
1. Springborn Laboratories 47
1. EVA Testing, 51
2. Module Exposure, 51
C. GLASS-REINFORCED CONCRETE 54
D. ELECTROSTATIC BONDING 55
C. POTTANTS 57
E. ELECTRICAL ISOLATION , 58
F. ANTI-SOILING 58
G. THIN-FILM PHOTOVOLTAICS 58
x
IX. REFERENCES .. 59
Figures
11. Outdoor Soiling Behavior of Tedlar 100BG30UT Plastic Film, With and Without a
Fluorocarbon Anti-soiling Coating . 20
12. Outdoor Soiling Behavior of Acrylar X-22417 Plastic Film, With and Without a
Fluorocarbon Anti-soiling Coating .. 21
13. Outdoor Soiling Behavior of Glass With and Without a Fluorocarbon Anti-soiling Coating 21
19. Effect of Air Temperatures (TA) on the Heat-Dissipation Behavior of a Glass-Superstrate Module. 30
22. LlT Versus Insolation for Sensor Technology, Inc. (Photowatt International, Inc.)
Block I Module 32
25. LlT Versus Insolation for Solar Power Corp. Block I Module . 33
xi
26. Structural Analysis, Deflection, and Thermal Stress. 34
33. Log-Log Plots of Computer Traces Given in Figure 28 for a Glass-Superstrate Design. 38
34. Horizontally Shifted Computer Traces of Figure 33 for Glass Using Reduced Variable, tiE . 38
38. Predicted Stress of Encapsulated Silicon Solar Cells Resulting from Thermal-Expansion
Differences in an Aluminum Substrate Module for a AT of 100°C, Using the Thermal
Stress Master Curve 40
40. Predicted Stresses in Encapsulated Silicon Solar Cells Resulting from Hygroscopic Expansion
of a Hardboard Panel from 0 % to 100 % Relative Humidity, Using the Thermal Stress
Master Curve 41
45. Illustrative Representation of the Natural Outdoor Aging Pattern for Unstabilized Polypropylene 50
46. Natural Outdoor Aging of Polypropylene on the Outdoor Heating Racks at 70,90, and 110°C. 51
xii
Tables
6. Cure of Elvax 150 EVA at Various Times and Temperatures with Lupersol-101 and
Lupersol-TBEC as Monitored by Gel Content in wt % 8
14. Effect of Hydrothermal Aging in Water at 40, 60, and 80°C on TBEC-Cured EVA with
30 vol % Glass Beads, With and Without Primer 14
15. Effect of Drying the Hydrothermally Aged Test Specimens Reported in Table 14 to
Assess the Reversible Recovery of Properties ... 14
16. JPL Soiling Data: Reduction Percentages in PV Cell Short-Circuit Current from
Soiling Layers on Cell Cover . 18
21. Experimental and Predicted NOCT Values for Spectrolab, Inc., and Solar Power Corp.
Block II Modules 31
25. Properties of Cured A-9918 EVA as a Function of Exposure Time to RS/4 UV at 55°C. 48
xiii
SECTION I
Introduction
In 1975, the encapsulation requirements for (1) Module life of 30 years (increased from 20
photovoltaic (PV) cells in terrestrial use had not been years).
comprehensively assessed, needs had not been defined,
nor potential low-cost encapsulation materials investi- (2) Encapsulation materials costs, including
gated. Therefore, the Flat-Plate Solar Array (FSA) Pro- substrate and/or superstrate not exceeding
ject set out to assess the needs to identify, and/or $1 .30/ft2 (1980 dollars).
develop new materials and new material technologies
that were inexpensive and could reliably protect and (3) Initial optical transmission of 90% and a loss of
support the fragile solar cells for years. The assess- less than 5% after 20 years of use.
ment led to the establishment of encapsulation needs,
goals, and detail requirements and a plan on how to (4) Capability to withstand an electrical breakdown
meet the goals. voltage of 3000 V DC.
The four basic functions of a PV module encapsula- (5) Structural integrity and durability to withstand
tion system are to: handling and weather.
(1) Provide structural support and positioning for (6) Cost-effective processing in an automated
the solar cell circuit assembly during fabrication, factory with high yields.
handling, storage, transportation, installation,
and operation in the terrestrial environment.
C, BACKGROUND
(2) Achieve and maintain maximum optical coupl-
ing between the silicon solar cell and the inci- The FSA Project goal established in 1975 was to
dent solar irradiation in a prescribed spectral develop the technology for manufacturing solar cell
region. modules at a manufactured price of $0.50 per watt, or
$5/ft2 (in 1975 dollars), assuming a 10% module effi-
(3) Achieve and maintain reliable electrical isolation ciency at an insolation level of 100 W/ft2, The module
of the solar cell circuit elements from both the was to have the necessary design features, sufficient
operational and safety viewpoints during the strength, and the appropriate materials of construction
useful life of the module. Voltage potentials to ensure outdoor operating performance for at least
above ground may exceed 1000 V. 20 years,
(4) Provide and maintain physical isolation of the Out of this $5/ft 2 cost goal, the initial allocation for
silicon solar cells and circuit components from all of the encapsulation construction materials was
exposure to hazardous or degrading environ- $0,25/ft2 (in May 1976), including structural support.
mental factors (e.g., hail, salt spray, and birds). The allocation was increased in 1979 to $14/m 2 , or
about $1 .30/ft2, to include also any required frame
and edge seals 1. The durability goal was increased in
A. TASK OBJECTIVES 1983 to 30 years.
To fulfill these functions most economically and Encapsulation material systems are defined as all
reliably, the objectives of the FSA Encapsulation Task construction materials required in a PV module to pro-
were to define, develop, demonstrate, and qualify vide mechanical support, electrical insulation, and envi-
encapsulation systems and materials and processes that ronmental isolation for the solar cells and their ancillary
meet the FSA Project module life, cost, and perform- electrical circuitry. Figure 1 illustrates early 1970 encap-
ance goals; and to develop and validate an encapsula- sulation systems employed by the emerging PV indus-
tion system life-prediction methodology based on model- try. In general, the encapsulation approaches were an
ing life-limiting failure modes and on conducting and extension of the technology for conformally coated elec-
analyzing aging tests. tronic circuit boards used for spacecraft and aircraft. The
circuit board panels were either a Nema-G10 fiber rein-
forced epoxy board or a glass fiber reinforced polyester
B. TASK GOALS panel, on one surface of which were mounted the
various electronic components which were then encap-
The goals of the FSA Encapsulation Task were to sulated in a conformal coating of a curable silicone elas-
develop encapsulation system technology adequate tomer. Early PV module designs also included the use
for:
METAL SUBSTRATES
(1) Delamination of the silicone elastomer from the
solar cells and substrate panels.
2
SECTION II
·· ABRASION RESISTANT
ANTIREFLECTIVE
mechanically supported. In the substrate design, the
cells are encapsulated in an elastomeric media on a
"
<J)
<l:
(:J
FRONT COVER
·· UV SCREENING
STRUCTURAL SUPERSTRATE
SILICONE OR rl - - - - - - - - - - 1 [ - - GLASS
POLYVINYL BUTYRAL - - - : ........ ........ ........ ........ I
I !-- GLASS OR B. ENCAPSULATION MATERIAL
PLASTIC FILM DEVELOPMENT
3
volume cost ($/ft 3) of each material, a comparative For expected temperature levels in operating
cost per ft2 of panel area was generated. The mini- modules (see References 66 and 67), "" 55°C in a
mum cost in 1976 dollars for each of the above panel rack-mounted array and possibly up to 80°C on a roof-
classifications, and associated material whether prac- top, there are three transparent polymers that are gener-
tical or not, is shown in Table 1. ally resistant to the weathering actions of ultraviolet (UV)
photooxidation, thermal oxidation, and hydrolysis. They
Note that onl~ wood and paper panel products cost are silicones as a general class of elastomeric materials,
less than $0.25/ft , thus, suggesting that only a sub- and two specific plastics, polymethyl methacrylate
strate design would meet the cost goals. For a super- (PMMA), and TEFLON FEP (Du Pont), a fluorocarbon
strate design, glass as a transparent structural material (see References 17, 68, and 69). Only silicones, which
was significantly lower costing than any commercially were expensive, had been available as low-modulus
available transparent structural plastics. For reasons of elastomers suitable for pottant application.
cost as well as a well-established history of outdoor
weatherability, glass was clearly selected for the super- All other transparent, low-modulus elastomers are,
strate designs, and transparent plastics for this use were in general, sensitive to some degree of weathering
not investigated under the JPL program. degradation. However, less weatherable and lower-eost
materials can be considered for pottant application if
For the investigation of substrate designs, both mild the module design can provide the necessary degree
steel and fiberboard (also called hardboard) were of environmental protection. For example, a hermetic
selected for investigation. It was recognized, however, design, such as a glass superstrate with a metal foil
that some development would be needed to achieve back cover and appropriate edge sealing, will essen-
long-term and low-eosting weatherproofing techniques tially isolate the interior pottant from exposure to oxygen
for the wood panel, and to achieve low-eosting methods and water vapor, with the glass itself prOViding a level of
of corrosion protection for the mild steel. UV shielding.
4
Table 2. Specifications and Requirements for Compounded Pottant Materials
Fabrication temperature ::; 170°C for either lamination or liquid pottant systems
weathering relationship as summarized in Table 3. In In the price range between $0.55 to $1 .50/lb,
general, materials costing less than $0.55/Ib were several transparent materials were found that were
sensitive to all three forms of the weathering actions. resistant to thermal oxidation and hydrolysis up to
Materials above $1 .50/1b could be found, such as the 80°C, but that were UV sensitive. None of the materials
silicones that were generally weatherable, but, for were liquid casting systems. Out of this subset of
other practical reasons, were not suitable for pottant materials, one polymer, ethylene vinyl acetate (EVA) at
application. $0.65/Ib stood out as the most viable pottant candidate,
on the basis of weatherability considerations, elasto-
Table 3. Transparent Polymeric Pottants meric properties, and module processing by dry-film
lamination techniques.
Modes of outdoor weathering degradation
Consequently, from these initial surveys, four pottant
Thermal oxidation materials emerged as most viable and are currently in
Hydrolysis various stages of development or industrial use. The
four pottants are based on EVA, ethylene methyl acry-
UV photooxidation late (EMA), poly-n-butyl acrylate (P-n-BA), and aliphatic
polyether urethane (PU). EVA and EMA are dry films
Costa/weathering relationship designed for vacuum-bag lamination at temperatures up
to 150°C. Above 120 °C during the lamination process,
> $1 .50/Ib: generally weatherable
EVA and EMA undergo peroxide crosslinking to tough,
$0.55 to $1.50/lb: UV sensitive, resistant to thermal rubbery thermosets. P-n-BA and PU are liquid casting
oxidation/hydrolysis systems. P-n-BA, a polymer/monomer syrup, was
developed jointly by JPL and Springborn Laboratories,
< $0.55/Ib: generally unweatherable and was formulated to cure within 15 min at 60°C. The
most promising PU for PV module application is a
aCost in 1976 dollars; the lowest costing trans- urethane designated Z-2591 , marketed by Development
parent polymer averaged around $0.23/Ib. Associates, North Kingston, Rhode Island.
5
a. Ethylene Vinyl Acetate. EVA is a copolymer Compounding
of ethylene and vinyl acetate typically sold in pellet Ingredients Problem
form by Du Pont and U.S. Industrial Chemicals, Inc.
(USI). The Du Pont name is Elvax; the USI trade name Peroxide Curing Agent
is Vynathane. the cost of EVA typically ranged between EVA immiscibility
$0.60 to $0.70/lb. in 1980 dollars. All commercially avail-
able grades of EVA were examined and the list reduced Rapid physical loss
to four candidates based on maximum transparency:
Elvax 150, Elvax 250, Elvax 4320, and Elvax 4355. Lupersol 101
Limited storage life
Because EVA is thermoplastic, processing into a module
is best accomplished by vacuum-bag lamination with a EVA cure problems
film of EVA. Therefore, based on film extrudability and
transparency, the best choice became Elvax 150. Weathering Stabilizers
Low molecular weight
As sold, Elvax 150 softens to a viscous melt above
70°C and, therefore, must be cured (cross-linked) for Tinuvin 770 Rapid physical depletion
service temperatures above 70°C. A cure system was
developed for Elvax 150 that results in a temperature- Cyasorb UV-531 Gradual loss of EVA
stable elastomer. Elvax 150 was also compounded with weathering protection
an antioxidant and UV stabilizers, which improved its
weather stability, but did not affect its transparency. The Table 4. Formulation of the First Generation EVA
formulation of the first-generation encapsulation-grade Encapsulation Film, Designated A-9918
EVA is given in Table 4, which carries the Springborn
Labs designation A-9918. These ingredients are com- Composition
pounded into Elvax 150 pellets, followed by extrusion at Component (Part by
85°C to form a continuous film. The thickness of the Weight)
clear film was nominally 18 mils. The peroxide curing
system is inactive below 100°C, so that film extruded at EVA (Elvax 150, DuPont) 100.0
85°C undergoes no curing reaction. The extruded film Lupersol 101 (peroxide) 1.5
retains the basic thermoplasticity of the Elvax 150.
Therefore, during vacuum-bag lamination, the material Naugard-P (antioxidant) 0.2
will soften and process as a conventional laminating Tinuvin 770 (UV stabilizer) 0.1
resin. Properties of Elvax 150 and cured A-9918 EVA Cyasorb UV-531 (UV stabilizer) 0.3
are given in Table 5.
This EVA pottant (A-9918) was well received by To correct these problems, a new curing agent
manufacturers of PV modules who reported certain designated Lupersol-TBEC was identified by Spring-
advantages of EVA when compared to PVB, a lami- born Laboratories. This curing agent is completely
nating film material developed for manufacturing auto- miscible with Elvax 150, and virtually eliminated prob-
mobile safety glass and in common use within the PV lems associated with storage life and cure. In fact,
module industry. The reported advantages of EVA were: TBEC was found to cure EVA faster, and at lower
temperatures, as compared to Lupersol-1 01. The cure
(1) Lower cost. characteristics of those two peroxide curing agents
are compared in Table 6, using the gel content of the
(2) Better appearance. cured EVA as one measure of the level of cure. EVA
is considered adequately cured if its gel content
(3) Better clarity. exceeds 65 % .
(4) Elimination of cold storage (required for PVB). In addition, the low molecular weight UV screening
agent Cyasorb UV-531 has been replaced with a chemi-
(5) Dimensional stability. cally attachable UV screening agent UV-2098, and a
polymeric (UV-3346) hindered amine light stabilizer
(6) No need to use a pressure autoclave. (HALS) has replaced Tinuvin 770. HALS are essentially
antioxidants effective against UV photooxidation. This
(7) Good flow properties and volumetric fill. second-generation EVA is an advanced EVA formulation
designated 18170, and is detailed in Table 7. Note that
Although this first generation encapsulation-grade Naugard-P (see Table 4), an antioxidant effective against
EVA was favorably received by the industry, its status thermal oxidation has been eliminated, since experi-
was still considered to be experimental. With increasing mental testing has indicated that UV-3346 also fulfills the
industrial experience, as well as observations noted same function.
during accelerated aging tests of this EVA formulation,
several limitations involving the A-9918 compounding Initial experience with fabrication and module per-
ingredients became evident, as follows: formance testing with the EVA 18170 pottant has indi-
cated a significant improvement over EVA A-9918.
6
Table 5, Properties of Elvax 150 and Cured A-9918 EVAa
Young's modulus b 23°C 850 Iblin 2 890 Ib/in 2 ASTM D-638 (Springborn)
23°C 700 Ib/in 2 ASTM D-1708 (Du Pont)
Hardness Shore A, 10s 65-73 76-79 ASTM D-2240 (Du Pont, Springborn)
Shore D, 10s 24 ASTM D-2240 (Du Pont)
aSources' Property measurements made at Springborn Laboratories under FSA Contract 954527.
Property measurements made at Spectrolab, Inc., under FSA Contract 955567.
Property measurements made at the JPL's analytical test facilities.
Various Du Pont Technical Bulletins on Elvax resins.
Du Pont Technical Bulletin, "Elvax 150 Resin as a Solar Photovoltaic Module Pottant, Technical Guide,"
Polymer Products Department, Technical Services Laboratory, Wilmington, Delaware (June 1982).
blnitial slope of stress-strain curve.
c Refer to the Remarks column.
d Made by Du Pont.
e Made by JPL.
7
Table 6. Cure of Elvax 150 EVA at Various Times and Temperatures with Lupersol-101 and Lupersol-TBEC as
Monitored by Gel Content in wt %
1 0 2.1 28.8
2 1.0 4.1 29.5 74.2 0 0 73.4 81.5
5 11.8 21.1 73.0 81.2 0 60.3 83.7 88.6
10 1.0 23.5 63.2 82.6 92.7 0 75.0 88.2 91.6
15 2.3 59.3 88.3 0 85.0 90.2 93.5
20 60.0 78.3 92.7 93.0
30 3.4 68.2 82.7 92.2 92.6
60 32.1 80.6
Composition
Component Function (Parts by Weight) Remarks
8
Table 8. Formulation of EMA Encapsulation Film Table 9. Formulation of P-n-BA Casting Liquid
Composition Composition
Component (Part by Weight) Component (Part by Weight)
EMA (TD 938, Gulf Oil Co.) 100.0 n-Butyl acrylate (monomer) 60.00
d. Aliphatic Polyether Urethane. Almost all Excluding glass, the only commercially available
commercially available polyurethanes are of the transparent, UV-screening plastic films that have been
aromatic, polyester type, which are not favorable identified are polyvinyl fluoride (PVF) films, Tedlar
because of their tendency toward hydrolysis of the (Du Pont), and PMMA films, Acrylar (3M Co.).
ester groups, and UV degradation because of UV
absorption by the aromatic structure. Only a few a. Tedlar. Du Pont markets three 1-mil-thick,
aliphatic polyether urethanes have been identified, clear, Tedlar fluorocarbon UV-screening films. The
and one of the more promising for PV module appli- designation of these three films are:
cation is a urethane designated Z-2591 , marketed by
Development Associates, North Kingston, Rhode (1) Tedlar 100 AG 30 UT.
Island. This material is being used industrially, and is
available for further evaluation by module manufac- (2) Tedlar 100 BG 15 UT.
turers for module designs requiring a casting rather
than a laminating process. (3) Tedlar 100 BG 30 UT.
9
An initial difficulty with Tedlar was poor adhesion Table 10. Back Coversa (White-Pigmented
to EVA and EMA. This was corrected by the use of an Plastic Films)
all-acrylic contact adhesive that could be coated directly
onto one surface of Tedlar films. The adhesive coating, Tedlar 150 BL 30 WH, 1.5 mils (DuPont)
a Du Pont product designated 68040, is dry and non-
tacky at ambient conditions; thus, coated Tedlar can be Tedlar 400 BS 20 WH, 4.0 mils (DuPont)
readily unwound from supply rolls. Experimental testing
Scotchpar 10 CP white, 1.0 mils (3M Co.)
indicates that when the adhesive is heated during the
EVA and EMA lamination cycle, strong adhesive bond- Scotchpar 20 CP white, 2.0 mils (3M Co.)
ing develops between the pottants and the Tedlar films.
The thickness of the adhesive coating ranged between Korad 63000 white, 3.0 mils (Xcel Corp.)
0.3 and 0.4 mil. An additional concern with Tedlar is a
potential toward shrinkage during outdoor service on a aApproximate cost per mil of thickness:
module, which can lead to cracks. Experimental studies (1) Tedlar PVF films, ;::; $0.07lft 2 .
suggest that the Tedlar film should be thermally (2) Scotchpar polyester films, ;::; $0.02/ft~
annealed before being fabricated into a module, to (3) Korad acrylic film, ;::; $0.02/ft 2 .
reduce this shrinkage concern.
10
SECTION III
One of the first encapsulation problems experienced In general, for glass, silane coupling agents have the
with PV modules fabricated in the early 1970s was chemical form X-{CH2)3-Si-(ORh where the functional
delamination of the silicone elastomers. At that time, PV group X is reactive with a polymer matrix, and can be
manufacturers were not employing adhesion promoters, varied depending on the polymer chemistry. The (OR)
and therefore, module interfaces in common with the groups are short chain ethers that hydrolyze in slightly
silicone materials were only in physical contact, and acidic water to produce alcohols that are reactive with
easily prone to separation if, for example, liquid water the surfaces of glass. The general scheme for chem-
were to penetrate to the interfaces. Delamination with ically bonding polymers to glass with silane coupling
silicone materials virtually vanished when adhesion agents is shown in Figure 5, and specifically for EVA,
promoters, recommended by silicone manufacturers, the X-group is a methacrylate. From studies of the inter-
were used. facial chemistry, there is growing evidence that coupl-
ing to a polymer involves not only chemical bonding,
With the decrease in use of silicone encapsulants, but also the formation of what is referred to as an
and the increase in use of hydrocarbon encapsulants "interpenetrating network" (IPN). An IPN is an inter-
such as EVA, the need arose for adhesion promoters phase that can be viewed as an interlocking of two
specifically developed for these new materials. The three-dimensional meshes. The current concept of a
adhesion promoters developed for EVA-type materials chemically bonded interface between EVA and glass
are based on organosilane chemistry, called silane is shown in Figure 6.
coupling agents. These coupling agents generate
primary chemical bonds across an interface, i.e., A. ADHESION PROMOTERS
chemical bonding (see References 16, 46, and 62). A
solution of an organosilane coupling agent in a solvent is Table 11 lists primers and adhesives that have been
called a primer solution, or just simply a primer. developed or identified for bonding EVA to glass,
Acrylar, Tedlar, and polyester plastic films. Experimental
The organosilane coupling agents are chemically quantities of the EVAlglass primer are available from
bifunctional molecules, and using glass and EVA as an Springborn Laboratories, under the designation A-11861 .
example, one of the chemical groups on a silane coupl- For bonding EVA to Tedlar, Du Pont identified an adhe-
ing agent is reactive with glass, while the other is reac- sive designated "68040" that is normally pre-coated as
tive with EVA. Thus, the interfacial bonding between a dry film on the Tedlar. Limited experimental testing
EVA and glass is accomplished by actual primary suggests that these same adhesion promoters work
chemical bonds bridging across their interface. equally well with EMA.
GENERAL STRUCTURE
X IS POLYMER REACTIVE GROUP
SUCH AS:
x x
I I • METHACRYLATE
TRIALKOXY RO - Si -OR HO- Si -OH • AMINE
SILANE I
I • EPOXY
OR OH • STYRENE
x x x
I I H, I
/' "
HO -Si -OH HO - Si - 0 0 - Si - OH
I I '" H ./ I
o o 0 x x x
:'H :'H :'H
, I I I
H H H Si - 0 - Si - 0 - Si - 0
I ! I I I I
o o o o 0 .0
I I
/// I I I
/7/77/7/l777/l/l777/////////o/
A//; - Si 0 Si - 0 - Si -
0-
. /;/;/////,//////////////////////////
11
CHEMICALLY AONDtD
INTERFACE INTERPtlASE With the exception of Tedlar, Table 12 lists adhesion
~ _----~A~-_--....., promoters for the Z-2591 polyurethane liqUid system
I :~:~xx>-<
that are available as commercial products from Develop-
ment Associates, North Kingston, Rhode Island, the
polyurethane manufacturer. An experimental primer for
bonding Z-2591 to Tedlar, developed by Dow Corning,
is also given in Table 12.
mILK
EVI\
0 - S'-
x;>-<>-< Table 12. Primer and Adhesives for Bonding
Z-2591 to Various Materials
I
0 - Si-
:~:~x~~;
PrimerlAdhesive
Material Designation a
Glasslmetals Z-3012
Composition,
Jabfe 11. Primer and Adhesives for Bonding Component wt%
EVA a to Various Materiafs
Formulation 1
Composition,
Component wt % Z-6020 silane 9.5
(Dow Corning)
Primer b for Bonding EVA to Glass
Water 0.5
Z-6030 silane (Dow Corning) 9.0
Methanol 90.0
Benzyl dimethyl amine 1.0
Lupersol 101 0.1 Formulation 2
Methanol 89.9 Z-6020 silane 2.5
(Dow Corning)
Primer for Bonding Ethylene Vinyl
Acetate to Polyester Films Z-6082 silane 2.5
(Dow Corning)
Z-6040 silane (Dow Corning) 1.25
Resimene 740 (Monsanto)c 23.75 Methanol 95.0
Isopropanol 75.0
a These adhesion promoters are marketed by
Primer for Bonding Ethylene Vinyl Acetate to Acrylar Development Associates, North Kingston, Rhode
Island, for their polyurethane product Z-2591.
Z-6020 (Dow Corning) 1.95 b Specially developed for mild steel; for other metals,
Z-6030 (Dow Corning) 0.05 use Z-3012.
Acroloid AT-51 36.0 c Both primers yield acceptable bond strengths on
(Rohm and Hass) freshly made test specimens, but long-term outdoor
Toluene 62.0 aging stability not known at this time, therefore,
these are still to be considered as experimental
Adhesive for Bonding EVA to Tedlar primer systems.
68040 Acrylic Contact Adhesive (Du Pont)
12
peeling of a i-in. strip of EVA from the flat glass sur- chemistry of a chemically bonded interface undergoing
face, and is expressed in units of pounds/(inch-of- hydrothermal aging. Therefore, a study was initiated at
width). The dry strength with unprimed glass is typically Case Western Reserve University, to directly interrogate
5 to 8 Ib/in., but will drop to zero when wet. a chemically bonded interface for chemical information,
using an infrared (IR) technique based on diffuse
Bond strengths measured with primed glass are reflectance from glass surfaces.
shown in Table 13, using the A-11861 EVA/glass primer
(see Table 11), for both Sunadex and window glass. An Using glass and EVA with primer A-11861 (see
experiment was also carried out with Sunadex glass Table 11), the concept was to monitor the IR spectra of
substituting the Lupersol-TBEC peroxide for Lupersol- the interface of test specimens undergoing hydrothermal
101, in the A-11861 EVA formuiation. Lastly, the primer aging. The initial specimens consisted of EVA laminated
was compounded into the EVA film to produce a self- onto flat glass that did not yield sufficient interfacial
priming EVA formulation. The results from this experi- surface area for IR detection. Therefore, it was decided
ment with window glass are also shown in Table 13. to disperse glass beads into EVA to increase the surface-
to-volume ratio. Experimentally, a loading of 30 vol %
glass beads provided sufficient surface for IR detection
In general, the dry bond strengths with L-101
of the interfacial chemistry. Both chemical interrogation
ranged from 35 to 39 Ib/in. for all experimental condi-
and mechanical property testing of hydrothermally aged
tions, and was 51.3 Ib/in. when using Lupersol-TBEC
in place of L-1 01. These values can be compared test specimens were carried out.
with 5 to 8 Ib/in. for unprimed glass.
Mechanical properties measured for primed and
unprimed EVA/glass specimens that were hydro-
The significant benefit of chemical bonding with the thermally aged at 40, 60, and 80°C for times up to
organosilane coupling agents is observed in the wet 2000 h are given in Table 14. Because it was noted
strengths, both after 2 weeks immersion in 25°C water, trial testing that these specimens tended to
and after 2 h in boiling water. Essentially the bond absorb large quantities of water, it was also decided
strengths are virtually unchanged. Without primer, EVA to measure and record the absorbed water content of
would have readily delaminated from the glass. the specimens, as given in Table 14. For comparison,
unfilled EVA under the same hydrothermal aging con-
There is industrial interest in having a commercially ditions absorbs less than 1 wt % water, therefore, the
available self-priming EVA to eliminate a separate enormous uptake of water associated with the filled
priming step currently used. Another observation with EVA was initially believed to be locally clustered at
important industrial ramifications is the indication that the glass/EVA interface. This was subsequently
the bond strength of the self-priming EVA formulation verified by IR.
improves with water exposure, reaching a level of
strength where, in mechanical peel testing, the EVA For the unprimed EVA/glass specimens, the general
fails rather than the bond (cohesive failure). pattern was to absorb copious quantities of water, such
as over 2000 wt % after 2000 h at 60°C, and to also
undergo a reduction in mechanical properties, at a rate
C. HYDROTHERMAL AGING that seems to increase with increasing temperature.
In general, the silanes employed for these coupling For the primed samples, there was a dramatic
agents are typically resistant to deterioration by UV light, decrease in the quantities of absorbed water, as com-
therefore, the main weathering concern is from the pared to the unprimed specimens, and overall, mechani-
potential of hydrothermal deterioration over a long cal properties were more generally preserved, even after
exposure time outdoors. Almost nothing is known about 500 and 1000 h at 80°C. For the primed sample aged
the chemical reactions, mechanisms, and kinetics of the 2000 h at 60 ac, its absorbed water content was only
Table 13. Adhesive Bond Strengths for EVA Bonded to Flat Glass
13
Table 14. EffecfJ of Hydrothermal Aging in Water at 40, 60, and ao°c on TBEC-Cured
EVA With 30 vol % Glass Beads, With and Without Primer
At 40°C At 60°C At 80°C
Characteristics Control
100 h 500 h 2000 h 100 h 250 h 500 h 2000 h 100 h 250 h 500 h 1000 h
Unprimed
Modulus, psi 2830 1600 1700 830 1800 1600 380 30 1700 930 1100 b
Ultimate strength, psi 1380 1295 1230 1210 1240 960 530 50 460 285 310 b
Ultimate elongation, % 600 600 570 510 570 515 300 60 365 120 100 b
Ll. absorbed water, %c 0,15 189 51,0 0,92 293 410 2015 17.4 568 503 b
Primed
Modulus, psi 2500 1600 1800 2400 2000 2500 2000 1600 2200 2300 3000 1700
Ultimate streflgth, psi 905 1070 900 1150 935 930 990 830 987 1010 910 725
Ultimate elongation, % 350 385 190 390 445 285 315 120 275 245 220 140
Ll. absorbed water, %c 019 2,0 35 036 4,0 6,3 34.7 10 12,9 16,9 61,7
measured ifl air at flominal conditions of 25°C (7rF) and 40% RH,
b Samples deteriorated,
cValues quoted are the percent grain in absorbed water compared to initial weight equilibrated at nominal room conditions of 25°C (7rF) and 40% RH,
34.7 wt %, as compared to its unprimed counterpart demonstrated that organosilane coupling agents are
with over 2000 wt %. These data in Table 14 clearly enormously effective in reducing the accumulation of in-
reveal the positive influence of the organosilane coupling terfacial water. This suggests that organosilanes
agents in both minimization of interfacial water absorb- chemically reacted onto metallic surfaces may function
tion, and preservation of mechanical properties. Dupli- as anti-eorrosion agents, through the action of reducing
cates of the specimens, whose properties are reported or excluding liquid water at the metallic surface.
in Table 14, were dried for 72 h in an air-eirculated
oven at 105°C, to dry out the specimens. The intent Several solar cell manufacturers deposit an alumi-
was to measure the mechanical properties after dry- num layer on the back surface of their solar cells, for
out to assess reversibility and recovery behavior of reasons related to enhanced performance. In the natural
the mechanical properties. These data are reported in environment, the surface of aluminum will oxidize, and
Table 15. In general, for all hydrothermal aging condi- then in a moist condition, this oxidized aluminum layer
tions, the dried out primed specimens virtually will form a hydrate called pseudoboehmite (see Refer-
recovered their initial properties. ence 72). This aluminum salt is friable, crumbly, and
very weak cohesively. In adhesively bonding polymers
to aluminum, it is this aluminum salt, when formed, that
D. CORROSION PREVENTION leads to low strength bonding failures. It was, therefore,
of interest to determine if the EVA/glass coupling agent,
Metallic corrosion usually requires the presence of Z-6030 (see Table 11), would also couple to aluminum,
liquid water, as one necessary requirement. The adhe- and stop the formation of pseudoboehmite in moist
sion studies with primed EVA/glass specimens clearly environments.
Table 15. Effect of Drying a the Hydrothermally Aged Test Specimens Reported
in Table 14 to Assess the Reversible Recovery of Properties
Unprimed
Modulus, psi 2830 2400 3400 2200 2500 3000 2700 1100 2700 1800 b b
Ultimate strength, psi 1380 1205 1280 1210 1270 1340 960 480 1175 560 b b
Ultimate elongation, % 600 590 565 570 590 575 500 280 545 290 b b
Primed
Modulus, psi 2500 2200 3000 2600 2600 3100 3100 2800 2600 2600 2800 2600
Ultimate strength, psi 905 965 990 860 955 890 980 850 905 920 1020 880
Ultimate elongation, % 350 325 315 260 290 240 240 240 240 310 405 260
aDried in circulating air oven for 72 h at 105°C,
unusable,
14
This work, carried out at the University of Cincin- from an aluminum surface can be observed through
nati, first observed that pseudoboehmite will generate a an overcoating of an EVA film.
characteristic IR band at 1080 cm- 1, which can be
detected on the aluminum surface using an IR technique The aluminized back surface of the same commer-
based on surface reflection. Figure 7 consists of plots cial solar cell was next primed with A-11861 primer,
of the IR spectra of an aluminized back surface of a followed by overcoating with a laminated and cured
commercial solar cell, when new (spectrum A), and layer of EVA. This work is current, but initial results
after 20 min (spectrum B), and 50 min (spectrum C) show that no pseudoboehmite has formed after i-week
immersion in boiling water. At 50 min, there was clear immersion in boiling water. The aging is continuing, but
IR evidence for the formation of pseudoboehmite. For- already the evidence is accumulating that a self-priming
tunately, EVA does not have any IR absorbance at EVA with Z-6030 coupling agent accomplishes both
this wavenumber, and thus this IR band originating structural bonding to glass, and also corrosion protection
for the solar cell's aluminized back surface.
z
o
U5
(f)
:2:
(f)
z
<t
0:
f-
a:
WAVENUMBER, em- 1
Figure 7. Reflection Infrared Spectra Obtained from the Aluminized Back Side of a Silicon Cell
that has been Coated with a Thin Film of EVA
15
SECTION
Anti-Soiling Coatings
Soil accumulating on the front surfaces of PV by rain. It was further observed that the rain-resistant
modules reduces light transmission to the solar cells base layer formed in about 30 to 60 days.
and, therefore, reduces power output. A study was
carried out to investigate experimentally the details of Closer examination of the base layer revealed that
natural soil accumulation, and derive a basic under- it, in turn, consisted of two layers (see Reference 55).
standing of the surface requirements to reduce soil Thus, including the top rain-removable layer, a total of
accumulation. The studies resulted in the develop- three soil layers, labeled A, B, and C from the module
ment of effective low-soiling surface treatments for surface out, could be identified. The three soiling layers
glass and plastic films. A summary of the experimental are described in Figure 9. Layer C is the rain-removable
studies, and performance of the low-soiling coatings top layer. Layer B is the soiling removable by washing.
are described in the following paragraphs. Layer A soil involves strong chemical attachment or
chemisoption and if formed on the module cover, it can
only be removed by abrasive scrubbing. Accepting the
A. NATURAL SOILING BEHAVIOR soiling layer concept, the soiling curves for each of the
seven transparent materials could be separated into light
The behavior of natural soiling on different material obscuration caused by the base layers (A + B), and by
surfaces was monitored at seven differe9t climatic the rain-removable Layer C. This was done by connect-
locations by measuring the reduction in the short cir- ing the data minimums in the asymptotic region of the
cuit current {Isd of solar cells positioned behind soiling curves, and assigning the magnitude of the dotted
seven transparent materials, that ranged from soft line to the layers" A + B" (they cannot be resolved
silicone elastomers to semi-hard plastic films, and hard by this data technique). For layer C, the difference
glasses (see References 73 and 74). In general, the soil- between the dotted line and the maximum observed
ing accumulation behaved as illustrated in Figure 8 by peak was calculated as the light obscuration caused
the solid oscillating line, indicating that some soil accu- by layer C. The light obscuration values assigned to
mulating during dry periods was removed during rain the layer "A + B", and layer C are given in Table 16,
periods. It was noted that the minimums of the oscilla- for the seven locations and seven materials.
tions could be connected by a dotted line, suggesting
the formation of a rain-resistant base layer of soil The data indicate that the largest quantity of rain-
deposited on the surface. On top of this rain-resistant resistant soil (Column A + B) was found on the soft sili-
soil layer resided a loose layer that could be removed cone, followed next by the semihard silicone and, least,
o~
f-
Z
w
0:
0:
:::J
U
f-
:::J
c
U
0:
U
~
0: ..,--- --------
o
I
(f)
....J ' 3 0 - 6 0 days
....J
W A + B
U
0:
«
....J
o(f) /
0"---------------------------------
~ o
(f) TIME, ma
(f)
o....J
17
LAYER C: TOP, LOOSE SURFACE OF SOIL,
REMOVED EASILY BY RAIN
by the remaining five harder materials. Although the for some combinations of sites and materials, that neither
numbers for these five materials are small, there is an layer A nor layer B has formed (the minima of the
indicated ranking. Comparing the plastic films, the fluoro- soiling curves were zero). The data suggest that the
carbon (Tedlar) is slightly better than the acrylic (Korad). formation of the rain-resistant soil layers are both
Comparing the glasses, the ranking (in improving order) material- and site-dependent, but that module cover
is soda-lime, aluminosilicate, and borosilicate. Note, that material dependency dominates.
Soft silicone 20 10 NA NA NA NA NA NA 25 8 24 6 24 7
RTV 615
Semihard silicone, 14 8 5 2 6 2 3 17 15 16 12 15 8
Q1-2577
Acrylic film, 3 8 o 8 2 2 5 14 3 13 3 11
Korad 212
Fluorocarbon 8 o 5 o 2 o 2 3 13 16 2 12
film, Tedlar
Soda-lime glass 2 6 4 2 2 o 2 3 9 4 12 3 9
Aluminosilicate 12 5 o 2 o 2 2 12 2 13 2 11
glass
Borosilicate o 7 o 5 o 2 o 2 11 15 13
glass
18
The site dependency of layer C relates to the atmo- From the soiling data, it is possible to theoretically
spheric concentrations of soiling materials, their types, (see References 22 and 55) derive seven characteristics
and rain cycles. The average of the six or seven required to have low-soiling surfaces, as follows:
values of light obscuration by layer C is also included
in Table 16. If the average value for layer C is treated (1) Hardness.
as a measure of the soiling characteristics of an
environment, then (of the sites listed in Table 16) JPL (2) Smoothness.
and Pasadena are the dirtiest, and Goldstone and
Table Mountain are the cleanest. (3) Hydrophobicity.
The soiling data from two JPL sites designated as (4) Low-surface energy.
34 and 45 deg, differing only in the module tilt angles,
show a reduction in layer C soil accumulation with (5) Nonstickiness (chemically clean of sticky
increasing tilt toward the vertical. Figure 10 is a plot materials, surface, and bulk).
of tilt-angle data for layer C. A linear extrapolation
based on two points would suggest no layer C (6) Cleanliness (chemically clean of water-
deposition at a near-vertical alignment. soluble salts and first-period elements,
surface, and bulk).
Or----------------,.--,
I (7) Weatherability (resistance to UV photo-
2
I
/ oxidation and/or hydrolysis).
I
4 I These seven postulates relate to surface require-
/ ments for resisting the formation of rain-resistant soil
o
UJ 6 I layers and, taken in total, lead to the conclusion that the
I- I
<1:
/ top surface must be a hard, smooth, fluorocarbon
U /
oU"J 8 material, or a very thin (micrometers) coating of a
U"J
I fluorocarbon on a hard, smooth backing (i.e., glass).
<{ ,~~ I
10 These requirements were guidelines for the develop-
/
UJ'
U U ment of durable fluorocarbon coatings to be applied to
za::
<1:w
r->- 12 solar-module surfaces to achieve low maintenance costs
1-<1:
~-' and to preserve high transparency.
U"J:::!
zo 14
/
<1:U"J Candidate materials for the outer surfaces of PV
a:: I I
1-1-
16 r- I
I modules currently consist of low-iron glass, Tedlar
'I~
,,~
fluorocarbon film (Du Pont Co., 100BG30UT), and a
S2
-' biaxially oriented acrylic film, Acrylar (3M Corp.,
u..
o /1/ X-22417). These materials are all relatively hard,
U"J
U"J smooth, and free of water-soluble residues. Experiments
o-' 20[ I I were conducted to determine if an improvement in soil-
19
three outer-surface candidate materials, and allowed to C. PERFORMANCE OF LOW-SOILING
dry in air and react chemically for 24 h. The treated COATINGS
materials were then soaked in water, simulating rain, to
determine if they were adequately attached chemically. The degree of soiling on the test specimens was
The criterion for judging this attachment was whether monitored by measurement of the percentage of
water would bead, or spread on the surfaces when the decrease in the Isc output of a standard silicon solar cell
materials were periodically removed from the water positioned behind the soiled test specimens. For Tedlar,
bath. By this criterion, both coatings were judged to the best coating is found to be E-3820, and Figure 11
have become permanently attached to glass, the E-3820 compares the soiling behavior of uncoated Tedlar (con-
to have become attached only to the Tedlar, and the trol) and E-3820-coated Tedlar. For Acrylar, the best
L-1668 to have become attached only to the Acrylar. coating is found to be E-3820 in combination with
ozone, and Figure 12 compares the soiling behavior of
To promote chemical attachment of the L-1668 on uncoated Acrylar (control) and the E-3820 ozone-coated
Tedlar, and the E-3820 on Acrylar, the surfaces of both Acrylar specimen. For glass, little difference is noted in
of these films were then activated by exposure to comparing E-3820 and L-1668, but E-3820 might be
ozone, to generate surface polar groups that would react slightly better (Figure 13).
chemically with the silanes, followed by brushing on the
fluorocarbon coating solutions. This technique worked
Comparing the uncoated controls, glass has the
well. Therefore, as an additional experiment effort,
least tendency to retain natural soil, followed by Tedlar
E-3820 was also applied to an ozone-treated Tedlar sur-
and then Acrylar. However, with the fluorocarbon anti-
face, and L-1668 was also applied to an ozone-treated
soiling coatings, the soiling behavior and optical losses
Acrylar surface. The concept was that the ozone treat-
for all three materials become essentially the same.
ment may also enhance the adhesion of these fluorocar-
bon coatings to the plastic films. Glass and plastic films
coated with the fluorocarbon coatings were then The soiling data were averaged over the 28-month
mOl:lnted on outdoor racks on the roof of Springborn period, and the summary of time-averaged values are
Laboratories' facilities in Enfield, Connecticut. Evaluation given in Table 17. The untreated glass control sample
was performed monthly and a record of rainfall was kept realized an average optical loss of about 2.65 % over
to correlate soiling effects with precipitation. The sur- the 28-month period; the Tedlar control realized an
faces of these test specimens were not washed or average loss of about 5.38 %; and the Acrylar control
touched with the hands. specimen realized an average loss of about 7.20 %.
orr----,,--.---.--,---,---.---,---,---,---,-Ir-.--,---.----,
\,
\
\
,,--//--,
, , \
/\/ \\\
\
.......
I
r--.. . . .
2 \ / \
\ 1',,,.. _..J "-
\ 1 '..... . . . . ...1
\ I E-3820
#
\1 SURFACE COATING
U 4
-'"
~
UJ
Ul
«
UJ
5
UJ
6
o
2 4 6 8 10 12 14 16 18 20 22 24 26 28
EXPOSURE, months
Figure 11. Outdoor Soiling Behavior of Tedlar 100BG30UT Plastic Film, With and Without a
Fluorocarbon Anti-soiling Coating
20
1\
I \
I \
I \
, I
I
\ I
I
I \
\,..- ... "'"
\
r--- ............
I
J \ I
"' .... ... I
I
J \
\ I
I \
\ I
I
#. 4
"'I. /\
"I
V
\
\
I
! "\
\
" I
" 'JI
I
I '\ E-382010l0NE
u \
I " SURFACE COATING
~
'" "\,
UJ
(/)
«
UJ
0:: 6
U
UJ
o
10
11 L-_---!:_ _-L_ _-l-_ _L-_-.:L_ _--L_ _.L-_---l_ _-L_ _--L_ _.L-_---J_ _-L_ _...J
12 14 16 18 20 22 26 28
EXPOSURE, months
Figure 12. Outdoor Soiling Behavior of Acrylar X-22417 Plastic Film, With and Without a Fluorocarbon
Anti-soiling Coating
';R
0
_Vl
u
~ 4
w
(f)
«
w
a:
u
w
0
6 _0- E.3820
- - - CONTROL
8 L-_--'-_ _-.L1_ _--L_ _l...-_--L_ _--L_ _l...-_--L_ _--L_ _.L-_---J_ _--L_ _.L-_---l
024 6 10 12 14 16 18 20 22 24 26 28
EXPOSURE, months
Figure 13. Outdoor Soiling Behavior of Glass With and Without a Fluorocarbon Anti-soiling Coating
The data in Table 17 also indicate clearly that from 2.65% for untreated glass, to 1.55% for treated.
E-3820 is the better-performing fluorocarbon coating Similarly, on Tedlar, the E-3820 optical losses were
for all three materials. On glass, the E-3820 coating reduced from 5.38 to 1.70 %. Acrylar realized an
resulted in a reduction of soiling-related optical iosses average reduction in soiling loss from 7.20 % down to
21
Table 17. Time-A veraged Values of the in the rainfall totals generally correlate with the soiling
28-Month Soiling Data highs and lows shown in Figures 11 , 12, and 13. A sus-
tained dry period with little rain occurred during the
Time-Averaged fourth to the tenth month, with no rain at all in the eighth
Materials Optical Losses, % and ninth months. This resulted in the maximum accu-
Glass mulation of surface soiling observed from all test speci-
mens over the entire outdoor exposure period. After the
Control 2.65 ninth and tenth months, rainfall began to increase and
the surfaces became cleaner.
With E-3820 1.55
With L-1668 1.59 During this exposure period, the monthly rainfall
totals were accumulated from a fair number of rain-
Tedlar storms distributed throughout the month. In the twenty-
first month, however, an especially intense and heavy
Control 5.38
rainstorm of several days duration accounted for almost
With E-3820 1.70 all of the monthly total. As shown in Figures 11, 12, and
With L-1 668/olone 4.28 13, this intense rainstorm removed virtually all of the
measurable soil from the fluorocarbon-coated surfaces.
With L-1668 4.43
With E-3820/olone 4.68 With respect to Layer C behavior, if it can be
assumed that no rain-resistant layers (A or B) formed
Acrylar on the E-3820 coated glass specimen, then the time-
average value of 1.55 % (see Table 17) can be con-
Control 7.20 sidered as the time-average optical loss associated
With E-382010lone 2.59 with the cyclical deposition and rain removal of layer
C soil, in Enfield, Connecticut. The higher minimums
With L-1668 4.21
observed for Tedlar and Acrylar, respectively 1.70
With E-3820 4.44 and 2.59%, may indicate the formation of layers that
With L-1668/0lone 5.15 resist removal by gentle rainfall, but not by intense
rainfall. This suggests possibilities for further perform-
ance gains from the use of improved fluorocarbon
2.59%, or a performance gain of nearly 4.61 %. These coating materials. .
performance gains can be economically important to the
electrical power output of a PV module. This work suggests that maintenance-cleaning
techniques for hard surfaces should be developed for
Figure 14 is a plot of the rainfall pattern in Enfield, layer C, perhaps a low-pressure water spray (rain
Connecticut, over the soiling exposure period. The data simulation) during dry cycles.
are plotted as monthly rainfall totals in inches versus the
month in which the rainfall occurred. The highs and lows In conclusion, low-surface energy treatments based
on fluorosilane chemistry seem to be effective in retard-
11
10
~ 9
I
~ 8
::i
~
7
u..
~ 6
~
a:
>-
...J
5
I
~ 4
o
~ 3
JJASONDJFMAMJJASONDJFMAMJJAS
1981 I 1982 I 1983
22
ing the accumulation of dirt on the candidate outer transmission measured with a standard cell and light
surfaces of interest. The most effective soil retardant source. The removal of accumulated soil correlated well
treatments identified to date are: for Sunadex glass, with rainfall, but not with precipitation such as snowfall.
E-3820; for Acrylar, ozone activation followed by These 28 months of experimental effort provided sup-
E-3820; and for Tedlar, treatment with E-3820. port for the theoretically derived considerations for low-
soiling coatings, and a rationale for future activities for
After 28 months of outdoor exposure, the E-3820, improvements in fluorocarbon coating chemistry. Experi-
anti-soiling treatments resulted in potential performance mental evidence suggests that layer C is site-dependent,
gains from nearly 1 % for glass to 4 % for Acrylar in light rain-frequency dependent, and possibly tilt-dependent.
23
SECTION V
Encapsulation Engineering
25
(2) Incident solar flux in the wavelength region 2. Minimizing Module Temperature
of 0.4 JLm to 1.1 JLm should be transmitted
maximally to the solar cells. The optical Analysis has shown that:
properties and features affecting this trans-
mission are surface reflection ("'" 4 %), anti- (1 ) The relevant thermal properties of encapsu-
reflective (AR) coating on the solar cell, lation materials regulating module operating
absorption bands in the encapsulation temperature are IR emissivity of the front
materials, and index-of-refraction mismatch and back surface, solar absorption of the
at the interfaces. back surface, and thermal conductivity. Fig-
ure 15 illustrates the strong sensitivity of the
(3) Front-side transparent encapsulation mate- solar cell temperature on back-side emissiv-
rials should .have virtually flat transmission ity, for a glass-superstrate design.
(no absorption bands) in the wavelengths
from 0.4 JLm to 1.1 JLm, and an integrated 56
transmittance ~ 98 %, after correcting for
surface reflection losses of about 8 %. Low-
cost pottant candidates described in this 55
document have these optical properties.
Computer predictions of power output of 54
modules with 10 to 25 mils of EVA indi-
cated no penalty because of EVA thickness.
High-iron (Fe + + +) glass does have u
0 53
undesirable absorption in the wavelength w
region from 0.4 JLm to 1.1 JLm. 0.:
::J 52
f-
(4) AR coatings on silicon solar cells are a «0.:
necessity. The AR coating should be optic- w 51
a.
ally matched with the pottant, but being :2
w
optically matched with air is acceptable, f- 50
resulting in only a small power loss when ...J
...J
encapsulated. However, significant power w
u 49
loss occur in cells without any AR coating. r±
«
...J
(5) AR coatings on the module top cover sur- 0 48
(f)
face may be beneficial, if application cost
performance and durability can provide a
47
cost-benefit advantage. AR coatings on the
second surface of glass, that is, at the pot-
tant interface, tend to reduce transmission. 46
Glass superstrates with AR coatings on
both sides are not recommended.
0.2 0.6 1.0
(6) Computer analysis and experimental expo-
sures to normal-incident light on stippled
BACK-SIDE EMISSIVITY,EB
glass superstrates, either stipple-up or
stipple-down, found no optical effects, either
beneficial or detrimental. Figure 15. Solar-Cell Temperature Versus Back-Side
Emissivity for a Glass-Superstrate Design
(7) Matching indexes of refraction of adjacent
material layers are desirable, but if not
done, back-reflection losses for the combi- (2) In terms of these thermal properties,
nations of glass, plastic-film front covers, module operating temperature is primarily
and pottant materials being considered are regulated by the IR emissivity of the front
small because the index-of-refraction differ- and back surfaces, and to a lesser extent
ences for these various materials are small. by thermal conductivity. This leads to the
The best situation for mismatched index-of- finding that a glass superstrate design and
refraction is to have them increased in each wood substrate design may have nominally
layer from the surface layer inward toward the same solar cell operating temperatures.
the cells. The reverse, decreasing index-of-
refraction toward the cells, can result in (3) The dominant control on module operating
power losses. temperature, which can be exercised
26
through selection of encapsulation FRONT·SURFACE 111 TEMPERATURE = 'F
121 EMISSIVITY = £F
materials, involves the use of front- and
back-cover materials with maximum IR
emissivity (E). Transparent glass and plastic- FRONT·SIDE THERMAL
RESISTIVITY
n FRONT
film front covers have E values ranging LAYERS l: (UKI
between 0.85 to 0.90. Back-cover materials n
m BACK
K2. L2 l: IUKI
LAYERS m
(4) Lower module operating temperature is
K. THERMAL CONDUCTIVITY
aided by the use of fins on the substrates L. MATERIAL THICKNESS
Figure 16. Heat Dissipation Model Using the equations in Table 20 and the thermal
resistivity values in Table 18, the equilibrium relationship
The thermal model for back-side and front-side between Qc and Tc for a 1/8-in-thick glass-superstrate
material layers is shown in Figure 17. The encapsula- module was found, and is plotted in Figure 18. Two
tion material layers can be described by the sum of cases are shown: a field-array installation in which heat
27
Table 18. Thermal Resistivities
Watts-mils Thermal
Material IK),
°c L, mils Resistance, LlK
Table 19. Thermal-Resistivity Sums for The absorption and conversion to heat of incident
Glass-Superstrate and Wooden-Substrate sunlight that must be dissipated from a module, 0c'
Module Designs can be estimated from simple considerations when
modules are classified into three distinct absorption
ft2_ °c types (Figure 21). Simplifying assumptions for all three
Module Design Thermal Resistivity, module types are that incident sunlight is normal to
Watts
the module surfaces, and that surface back-reflection
L (LlK)front = 52.7 x 10- 3 losses are 4 % .
Glass, EVA,a
Mylar, Module Type A in Figure 21 has 100 % solar cell
L (LlK)back = 16.1 x 10- 3 area coverage. Hence, 0c is simply entering insolation
minus any electric power output: mathematically,
L (LlK)front = 15.4 x 10- 3
Acrylic, EVA,a 0c = (0.96) (S) - electric power (1)
Wood, Mylar
L (LlK)back = 194.1 x 10- 3 For normal operating cell temperature (NOCT)
estimates, electric power is zero; therefore,
aThis example assumes that the EVA layer in front 0c = (0.96) (S).
of the cells and the EVA layer behind the cells
are each 10 mils thick. Module Type B in Figure 21 has partial solar-eell
coverage (less than 100 %), with transparent intercell
spaces. Denoting the fraction of solar-eell area
dissipation can occur from both the front and back sur- coverage as A, then 0c is estimated by
faces, and a rooftop installation where heat dissipation
can occur from the front side only (I.e., 0B = 0). 0c = (0.96) (A) (S) - electrical power (2)
The relationship between Tc and qp
is nearly linear For this case, it is assumed that all insolation
over the range from 30 to 100 mW/cm . For the exam- entering the transparent intercell area, 1 - A, passes
ple shown in Figure 18, if the glass-superstrate module out of the module with no heat contribution. Again, for
mounted on an array installation were required to dissi- NOCT estimates, electric power is zero, and,
pate 70 mW/cm 2 of heat, its equilibrium temperature therefore, 0c = (0.96) (A) (S).
would be near 47°C. Mounted on a rooftop with no
back-side heat dissipation (OB = 0), its equilibrium tem-
Module Type C, in Figure 21, has partial solar cell
perature to dissipate 70 mW/cm 2 would be near 67°C.
coverage, with a white background in the intercell
spaces. Therefore, 0c is estimated:
The effect of air temperature on the operating
temperature of a glass-superstrate module is shown in
Figure 19 for air temperatures of 10, 20, and 30°C. 0c = (0.96) (A) (S) + (0.96) (1 - A)
(S)/3 - electrical power (3a)
Figure 20 compares the heat-dissipation behavior
of three module designs: a glass-superstrate module, a The assumption is made that one third of the sunlight
wooden-substrate module, and a mild-steel substrate incident on the white background is absorbed by the
module.
28
Table 20. Heat Transfer Equations for Heat-Dissipation Modeling
OFF = 0free convection= (1.52) (cos 1./;) 1/3 (t F - t A )4/3 (W)/(m 2) (0C)
0WF = 0wind convection = H (t F - t A )
4
0RF = 0radiation = (TF - Tsky4)
0FB = 0free convection = (1.31) (cos 1./;) 1/3 (t B - t A )4/3 (W)/(m 2) (0C)
0c = 0front + 0back
Terms and Constants
solar cells. Again, for NOCT estimates, electric power rooftop installations. The Spectrolab Block II module
is zero, and, therefore had a glass superstrate with transparent intercell area
(Module Type A); the Solar Power Block II module
0c = (0.96) (A) (S) + (0.96) (1 - A) (S)/3. (3b) had a structural-plastic substrate with a white back-
ground in the intercell area (Module Type C). Both
NOCT values have been measured experimentally module designs had a solar-cell area coverage A of
and reported (see Reference 66) for a Block II Spectro- about 0.7. The experimentally measured NOTC
lab, Inc., module in an array installation and for a values are given in Table 21.
Block II Solar Power Corp. module in both array and
29
90 AIR TEMPERATURE = 20°C (118 INCH· THICK GLASS)
WIND SPEED = 1 m/s ROOFTOP
TILT ANGLE = 34 deg INSTALLA TlON
SURFACE EMISSIVITIES: (OB = 0)
FRONT = 0.9
BACK = 0.9
80 GROUND EMISSIVITY = 0.8
u
°
U
f-
w 70
a:
;:)
f-
<t:
a:
w
a.
2 60 ARRAY
w
f- INSTALLATION
...J
...J
W
U
a:
<t:
...J 50
0
(f)
40
30 40 50 60 70 80 90 100
HEAT DISSIPATION 0c. mW/cm 2
AIR TEMPERATURE
Ta.oc
70
WIND SPEED = 1 m/s (118 INCH·THICK GLASS)
TILT ANGLE = 34 deg 1OOC} ROOFTOP
SURFACE EMISSIVITIES: 20°C INSTALLATION
FRONT = 0.9 30°C (OB = 0)
BACK = 0.9
GROUND EMISSIVITY"" 0.8
60
50
u
°
.;,
f-
40 10°C
20°C } ARRAY
I
u INSTALLATION
f- 30°C
30
20
30 40 50 60 70 80 90 100
HEAT DISSIPATION 0c. mW/cm 2
Figure 19. Effect of Air Temperatures (TA) on the Heat-Dissipation Behavior of a Glass-Superstrate Module
30
ROOFTOP
AIR TEMPERATURE ~ 20°C
WiND SPEED = 1 m/s
TILT ANGLE = 34 deg GLASS SUPERSTRATE
SURFACE EMISSIViTlES:
8(1 . FRONT = 09 WOOD AND STEEL
BACK = 0.9 SUBSTRATE
GROUND EMISSiViTY = 0.8
v
o
(.I
.... 70
w
a:
:::>
I-
<: ARRAY
a:
w
~ 60 WOODEN SUBSTRATE
w GLASS SUPERSTRATE
I-
-'
-'
W
';'1 STEEL SUBSTRATE
c::
::5 50
otfl
PANEL THICKNESSES
30 40 50 60 70 80 90 100
HEAT DISSIPATION 0C. mW/cm 2
QC = 10.96)IAIiSI + 0.96 S 11
---I
A1I3 - POWER OUTPUT
Solar Power
Array Installation 45 to 46°C 45°C
Solar Power
Rooftop Installation 61.5°C 60°C
Figure 21. Module Models for Q c Calculations
aReference 66.
31
46°e, The rooftop NOCT value is estimated to be their specific encapsulation systems, or encapsulation
near 60°C, using the same value of 0c and the design options, The heat dissipation curves shown in
wooden-substrate and steel-substrate rooftop heat- Figures 18, 19, and 20 were generated with the
dissipation line in Figure 20, equations in Table 20, Interestingly, these computer-
generated curves are virtually linear,
The method described herein for estimating module
operating temperatures is intended to be a convenient Figures 22 through 25 are reproduced from
desktop approach for PV module designers assessing Reference 77, and are plots of Tcell - Tair (0C) versus
50,"'1'"----------------------------------,
40
Figure 22. tl T Versus Insolation for Sensor Technology, Inc, (Photowatt International, Inc.) Block I Module
50--"1- - - - - - - - - - - - - - - - - - - - - - - - - - - - - ,
ALUMINUM SUBSTRATE DESIGN
40
1
,
i
u
°.n 30
E
co
I-
a:;
"
I- 20
10
T cell - T amb = iii S
O,-J
I-----------_----c----_---_---,__------l
70 75 80 85 90 95 100 105 110
INSOLATION S, mW;cm 2
32
5o.------------------------------------
EPOXY G 10 SUBSTRATE DESIGN
40
u
30
°
..c
E
.- <0
a;
.-u 20
T cell - T amb = mS
'0 m= 0.302, °c • cm 2 /mW
°7·l-:-10----7~5----8~0----85-----:-9·~0------:9.,-'5----10-r'O----,-cO-S---,-,.·:-O-.J
INSOLATION S, mW/cm 2
40
I
u
~ . ".;~ '~"-i~1I
; ~-i[ •• ;:,.~
° 30 ~ ,~~, ~ I ,,~: ,t -:J1"':;
..c .~
E
.- <0
a;
.-u 20
Tcell - Tamb = mS
10 m= 0.312, °C • cm 2 /mW
o~---_----::_--_:'------:"-----:-:=__---=__------~-'
70 75 80 85 90 95 100 105 "0
INSOLATION S, mW/cm 2
Figure 25. !1T Versus Insolation for Solar Power Corp., Block I Module
insolation in mW/cm 2 for four early 1970 substrate insolation level S. Thus, the computer-generated curves
design modules mounted outdoors at JPL in Pasadena, are in agreement with these experimental observations.
California. Experimental observations revealed that the What appears as data scatter around the straight lines in
temperature difference between the solar cell and ambi- Figures 22 through 25, was related to local fluctuations
ent air was essentially linearly proportional to the solar in ground wind velocities around the modules (see
Reference 75).
33
B. STRUCTURAL MODELING substrate design. The structural properties (Table 22) of
the glass, wood, and mild steel were fixed-input data.
Outdoors, a PV module is subjected to daily cycles The pottant was treated as a variable, expressed in
of thermal expansion and contraction associated with terms of its Young's modulus. Output data consisted of
the diurnal temperature swings, and to out-of-plane the stress distribution throughout the module (especially
deflections when subjected to wind. The stresses set up solar cell stresses), calculated as a function of pottant
throughout a PV module from differential expansions modulus, and pottant thickness between the cells and
and/or deflections must be limited to levels that would the structural panel. The structural analysis model is
not cause the solar cells to fracture or crack over the summarized in Figure 26.
outdoor service lifetime. The purpose of the structural
analysis was to understand the relation between encap- INPUT PROPERTIES ~Q.Q.uLE DESIGN£EATURtS
sulation designs, that is, tl)e mechanical and thermal MODULUS 1.2 X 1.2m SOUARE
TENSILE STRENGTH 10 x 10cm SOUARE CELLS
properties of the encapsulation materials, and the level THERMAL-EXPANSION COEFFICIENT 1.3-mm CELL SPACING
of stresses established in the solar cells when the PANEL THICKNESS
POTTANT
Thermal-Expansion Allowable
Modulus, Coefficient, Stress
Material Ib/in. 2 in.linPC klb/in. 2
Glass
Tempered 10 x 106 9.2 x 10- 6 13
Annealed 10 x 106 9.2 x 10- 6 1 to 3.6
Silicon (PV
34
18
16 : . . . . . . _ - - - - - - - - - - - - - - - - - - - - - - - - - E E 750, 000 16
N 14
~ .£ . 14
::J-- ALLOWABLE CELL STRESS IN DEFLECTION
~:9 12
-~--------- ::
--'"
Xen
«en
~w 10
a:
...JI-
~en 8
U...J
,« E = 2500
a: a..
« U - 6 6
...J
o~
en a:
a.. 4
~---------E = 1000
4
~----------E = 500
2 2
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
POTTANT THICKNESS, mils
9 9
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
POTTANT THICKNESS, mils
SILICON SOLAR-CELL DIMENSIONS = 4 X 4 X 0.01 O-in. SQUARE
Figure 28. Thermal Stress Analysis (ilT = 100 o e).· Glass-Superstrate Design
a temperature range of 100 °e. For deflection, assum- For pottant modulus of 1000 Ib/in. 2 , (i.e., EVA), the
ing that 8000 Ib/in. 2 is the allowable solar-cell stress solar-cell separation distance or pottant thickness
in deflection, the calculated stress curves indicate that must be > 4 to 5 mils, and the thickness must be
the solar cells must be separated from the glass by a about 11 mils for a pottant modulus of 2500 Ib/in. 2.
pottant thickness of at least 2 to 3 mils for a ~ottant Using a pottant with a modulus 50,000 Ib/in. 2 , the ten-
material having a Young's modulus of 500 Iblin. ,which sile stress in the solar cell will exceed 8000 Ib/in. 2 for
is typical of room-temperature vulcanized (RTV) silicones. any thickness.
35
For thermal-expansion stressing, (see Figure 28), plate of less than 0.028 in. thickness will experience a
the calculated relationship between solar-cell tensile peak bending stress exceeding its allowable stress limit
stress, pottant thickness, and modulus behave similarly of 28,000 Ib/in. 2.
as observed for the bending deflection analysis. Com-
paring Figure 28 for thermal stressing with Figure 27 for For a 1000-lb/in. 2 pottant modulus, the calculated
deflection stressing reveals that the deflection stressing relationship between solar-cell tensile stress and pottant
of the glass-superstrate module determines the minimum thickness for each of the three steel-plate thicknesses
pottant thickness required, assuming that the allowable are plotted in Figure 30. Because the level of the out-of-
cell stress in tension is 5000 Ib/in. 2. plane deflection at constant pressure loading increases
with decreasing plate thickness, the bending stresses
For this glass superstrate analysis, the level of imposed on the solar cells increase with decreasing
stresses developed in the glass were significantly less plate thickness. For a steel plate thickness of 0.028 in.,
than the stress-design lim!"t for tempered glass. approximately 12 mils of pottant are required between
the cells and the steel plate to have the deflection stress
2. Mild Steel Substrate Design in the cells less than 8,000 Ib/in. 2 , the allowable cell-
stress estimate in deflection.
In thermal stressing over a 100°C-temperature Increasing the plate thickness to about 0.087 in.
range, analysis indicated that the tensile stresses reduces the pottant thickness required to about 4 mils,
developed in the solar cells are independent of the which coincidentally matches the pottant thickness
thickness of the mild-steel panel. As in the glass- requirement for thermal stressing (see Figure 29). A
superstrate design, the tensile stresses in the cells are plate thickness of 0.168 in. is sufficiently stiff against a
a function of the pottant thickness and modulus. 50 Ib/ft 2 wind load that the bending stresses imposed
on the solar cells are below the allowable cell stress,
The calculated relationship between solar-cell ten- down to a pottant thickness of 1 mil.
sile stress and pottant thickness resulting from thermal
stressing is shown in Figure 29, for a single case of pot- 3. Wood-Hardboard Substrate Designs
tant with a Young's modulus equal to 1000 Ib/in. 2 (i.e.,
EVA). Assuming an allowable cell stress in tension of Hardboards are typically sold in thicknesses
5000 Ib/in. 2, the minimum pottant thickness required that incrementally increase by 1/8 in., such as 1/8, 1/4,
between the cells and the steel plate is about 4 mils. 3/8-in., etc. As with mild steel, stresses developed in
solar cells from thermal expansion differences are
The deflection analysis for 50-lb/ft 2 wind loading virtually independent of the thickness of the wood panel.
on an unribbed steel panel has been analyzed for a The results of the thermal stress analysis for hardboard
single pottant with a modulus of 1000 Ib/in. 2, and for is shown in Figure 31 . There is virtually little effect from
three different thicknesses of a steel plate: 0.168, 0.087, the pottant as the thermal expansion coefficients of
and 0.028 in. The peak bending stress developed in a hardboard and solar cells are very close. For deflection
4-tt2 steel plate of these thicknesses during wind loading (Figure 32), the minimum panel thickness allowable for a
of 50 Iblft 2 are 5000 Ib/in.2 for 0.168 in., 15,000 Ib/in.2 4-n 2 module is 1/4 in., but 1/8 in. may be allowable if
for 0.087 and 28,000 Ib/in. 2 for 0.028 in. A steel- structurally reinforced with stiffening ribs.
-- -- -- -- -- -- -- -- -- 5
2 345 6 7 8 9 10 11 12 13 14 15 16 17 18 19
POTTANT THICKNESS, mils
22
a = PEAK BENDING STRESS IN STEEL, Ib/in. 2
20 POTTANT MODULUS = 1000 Ib/in. 2 20
18 18
~:§
'" 16 16
:J::Q a = 28 klb/in. 2, 0.028 in. THICK
~.;,{.
X - 14 14
~i2 ALLOWABLE
~LlJ
0: 12 CELL STRESS IN 12
....If-
LtiC/)
U-l
DEFLECTION ~
,~ 10 10
0: a..
~-
-lU 8 ----
02 8
C/)cc
a..
6 6
a = 15 klb/in. 2 , 0.087 in. THICK
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
POTTANT THICKNESS, mils
'" .~
:0 9
u)
C/) 8 E = POTTANT MODULUS, Ib/in. 2
LlJ 8
0:
f-
C/) 7 7
....I
~ ALLOWABLE CELL STRESS E= 250,000
a.. 6 IN TENSION 6
U
2
0: 5
.L _ 5
a..
~
:J 4 4
~
X
~ 3
~
....I
....I 2
LlJ
U
ci::.
~
....I
0
C/)
2 4 5 6 7 8 12 13 14 15 16 17 18 19
37
N
c
c:::: 18 r--,--,----,--.-r--,---r----r-.--r-..,.--,---,--....---..-..,.--,---.,--..,.----, 18
.0
.:x.
(J = PEAK BENDING STRESS IN W02D, Ib/in.
2 16
(fJ 16
(fJ POTTANT MODULUS = 1000Ib/in.
LU
a: 14 14
I-
(fJ
ALLOWABLE CELL STRESS
-I
«0.. 12
_ ~. IN DEFLECTION _ u = 1600 Iblin. 2 • NO RIBS. 1/4·;n. THICK
12
U
z 10 10
a:
:2
0..
8
----------------~
_--~~
-===-------- ----- .. 8
::)
:2
x 6 500 Ib/in. 2 , RIBS, 11S-in. THICK 6
«
:2 4 4
....I
-I
LU
u 2 2
ri:
«-I
0 4 6 7 8 9 10 11 12 13 14 15 16 17 18 19
(f)
lOOr-' 100
~~ ~
i!:
Xv>
~~
lOr
POTTANT MODULUS E, klb/in. 2
~:;;
o
L GLASS PROPERTIES: ~~ I~:~ 2.5
f/1 MODULUS M " 10 x 106 lb/in. 2 1.0
THERMAL EXPANSION COEFFICIENT a • 9.2 x 10-6 0c- I 0.5
I. O::-:---'---L-"-'--'-'-~c----'----1--'--L.l-'-'-W-,---'--'---L.J--"-'...LW
0.1 1.0 10.0 100.0
0.1 1.0 lO
1. 0 :--:-_-'---L--'--'--W-~_----'---J..----'-...L-l-J....L.JJ,-_-,--,---1-l.-'-'-.uJ
100
POTTANT REOUCED VAR IABlE, tiE, milslklb/in. 2
t, POTTANT THICKNESS, mils
Figure 33. Log-Log Plots of Computer Traces Given this general finding, the master-eurve concept
Given in Figure 28 for a was expanded to include additional structural param-
G/ass-Superstrate Design eters relevant to the PV modules. The complete master
38
Examination of these predicted log-log traces indi-
cates that aluminum generates greater solar cell stresses
~ than glass, wood, or steel because of thermal-expansion
~
-:E
differences. The use of a pottant having a Young's
->'
V>
modulus of 1000 Ib/in. 2 , such as EVA, would require
~. that the pottant thickness be at least 14 mils, and higher-
'"
~ 10 modulus pottants would have to be used in correspond-
'"
;; ingly higher thicknesses. The dual handicap for alumi-
o 5 • SOLAR CEll MAXIMUM PRINCIPAL STRESS,
lj t • POITANTTIHCKNESS. mil num of a high thermal-expansion coefficient and higher
::> E • POTTANT MODULUS, klblln. 2 material cost as compared to glass, wood, and steel,
~ M • MODULUS OF STRUCTURAL PANEl, ~Iblln. 2
'" a • THERMAL EXPANSION COEFFICIENT OF STRUCTURAL PANEl, °C-1 makes it an unattractive module substrate-panel material.
6T· TEMPERATURE DIFFERENCE, °c
6. Wood Panel
1 J:;;-O-~~~'--'-'~JOO=-----'----'--'-..l-L..LJ..Jl:LOOO:---.L---'--'-.J.-LJOLJOOO~J
The hygroscopic-expansion coefficient for hard-
board is 50 x 10-6 in.lin. per one percent relative
REDUCED VAH IABIE, taM/E
10 3
Sc - SOLAR CELL MAX STRESS, klb/in. 2
SSP - STRUCTURAL PANEL MAX STRESS, klb/in. 2
ESp - MODULUS OF STRUCTURAL PANEL, Ib/in. 2
~ tsp - THICKNESS OF STRUCTURAL PANEL, in.
~
NOTE: USE CURVE A FOR PRESSURE,;; 10 LB/IN.; USE CURVE B FOR PRESSURE 50 LBIiN. 2
FOR INTERMEDIATE LOADS. INTERPOLATE LINEARLY BETWEEN A AND B
39
-
l.D
b
x
u
<:s
r--
'<t 0..
(/l
10- 2 C--------------------------.
I
~I
<:s
NOTE: + Sc - TENSION
-- ":
u
- Sc - COMPRESSION
...J
-
N
~
w
U
~ r-
~
<1 10- 3 Sc - SOLAR CELL MAXIMUM PRINCIPAL STRESS, klb/in. 2
2
U
(/l
-
C'l
~
0..
(/l
EC
tc
LC
-
-
-
MODULUS OF SOLAR CELL, Ib/in. 2
THICKNESS OF SOLAR CELL, mils
LENGTH OF SOLAR CELL, in.
w ac - THERMAL EXPANSION COEFFICIENT OF SOLAR CELL, 'C
0..
(/l tp - POTTANT THICKNESS, mils
<:l
Ep - POTTANT MODULUS, klb/in. 2
w
...J ESp - MODULUS OF STRUCTURAL PANEL, klb/in. 2
Cll
« asp - THERMAL EXPANSION COEFFICIENT OF STRUCTURAL PANEL, ·C-1
ii: ~T - TEMPERATURE DiFFERENCE, ·C
« 10-4 1....-_...I.--.L-.L....L-l.....LLU_ _..L--L-L...LLLUl_ _L-L-L-LLLu.J
>
cw 10 100 1000 10.000
u
::> REDUCED VARIABLE (t E)p (Ea)sp
cw
a:
_
M
'0
ALLOWABLE CELL
~z: ::~
:i
....
l?
§ , EQUILIBRATION
I TO ATMOSPHERE
I PRESSURE
20 40 60 BO 160
The predicted log-log traces for three levels of material to damp the generated expansion stresses
down to acceptable levels. A Qottant having a
Young's modulus of the pottant (E = 500, 1000, and
2000 Ib/in. 2) are plotted in Figure 40 for a relative- Young's modulus of 500 lb/in. 2 would have to be
humidity excursion of 100 % (i.e., dry wood to used at a predicted thickness of about 33 mils and a
saturated wood). pottant, such as EVA having a Young's mOdul~s of
near 1000 Ib/in. 2, would have to be at least 66 mils
thick. Even if the relative-humidity excursion after
The predicted solar-cell tensile stress response due vacuum-bag lamination were only up to 50 %, which
to humidity requires very thick layers of pottant is more realistic, the thickness of a pottant, such as
40
Ijy~G~ROSCOPIC STRESS ANALYSIS
(see References 52 and 56). This was part of an
overall program related to accelerated aging of
encapsulation pottant materials, and knowledge of
electrical stress levels to which pottants would be
subjected in service was sought.
WOOD PROPERTIES,
MODULUS M • 0.8 x 106 Iblin. 2
HYGROSCOPI C~EXPANS ION COEFflC IENT a· 50 x 10';; A mathematical analysis of the computer model
~:::::::::==:::::::::::.::::.:= of the electrical fields led to a possible fundamental
definition of an intrinsic dielectric strength of insulation
materials that could be stated as a basic material
property independent of any test technique, geometry,
or service environment. This is similar to other pure
material properties such as Young's modulus, index
of refraction, and coefficient of thermal expansion.
C. ELECTRICAL INSULATION (SAFETY) that has been historically observed (see Reference 80)
as a data correlation for eiectrical-breakdown voltage
The pottant in a PV module functions as an electri- as a function of sample thickness, which has been
cal insulation material, isolating the electrically-active explained as a material property. The computer result
solar cells and associated circuitry from external suggests that this behavior is a consequence of the
grounds and/or human contact. In series-connected thickness dependence of the electrical field, and that
PV module arrays, circuit voltages above ground may electrical breakdown occurs whenever a critical value
exceed 1000-V DC. Therefore, the long-term (about of Emax is reached or exceeded. This criticai value of
30 years) DC electrical insulation qualities of pottant Emax , herein designated S, is speculated to be the
materials such as EVA in outdoor weathering intrinsic dielectric strength of electrical insulation
environments is a major concern.
41
materials. The question then arises if Equation (7) is property. However, the similarity of Equation (8) with
unique to the geometry analzyed in Figure 41, or is it these series expanded expressions suggest that, at
a description of a general behavior for all electrode constant Emax , this is a manifestation of the thickness
geometries of whatever kind. dependence of the spatial distribution of the electrical
field. If true, then voltage breakdown of an insulation
Table 23 details two analytical solutions of material is occurring whenever a critical, but constant
Laplace's field equations for test electrode geometries value of the potential gradient Emax is reached or
consisting of a needle tip-to-ground plane configuration, exceeded on a electrode surface. It is suggested that
and for a needle tip-to-needle tip configuration (see this critical value of Emax may be the intrinsic dielec-
References 81 and 82). Both of these analytical solu- tric strength of an insulation material, hereafter desig-
tions can be series-expanded (see Reference 57), and
nated as S. From experimental voltage breakdown
when the first two terms of each of their respective
series expansions are algebraically combined, the values measured as a function of thickness t, the
result is: constants k, a, and n in Equation (8) can be derived
by least-squares techniques, and, therefore, for t == 0,
Tip-to-Tip S == K(a) - n that is identically equal to Emax (Equa-
tion 12) as K == Emax(a)n.
VA == Emax (3R) (t + 3R)-1 (9)
1. Voltage-Breakdown Data
Tip-to-Ground
Flat-Plate Solar Array testing related to dielec-
VA == Emax (R)2/3 (t + R)-2/3 (10) tric strength and voltage breakdown measurements of
encapsulation pottant materials is limited, and to date
only one preliminary set of alternating current (AC)
Table 23. Analytical Solutions of Laplace's Field voltage-breakdown data has been measured for the
Equations for Needle Electrodes EVA pottant. At the time of this preliminary test, it was
convenient to use a symmetric pairing of electrodes
Tip-to-Ground Plane (Reference 81): that was not dictated by any of the concepts or
theories being described. It turned out to be a for-
(dV/dY)max 2V At PILn (0)
tuitous choice. The test results measured on three
P == (1 + R/t) 112/R thicknesses of EVA film are given in Table 24, along
with the calculated average dielectric strength VA.
o == [2t + R + 2t 1/2 (t + R)1/2]/R
Table 24. Average AC Breakdown Voltage of EVA
Tip-to-Tip Plane (Reference 82):
for Three Film Thicknesses
Average Average
(dV/dY)max AC Dielectric
2R tanh -1 [t/(t + 2R)]1/2 Thickness Breakdown Strength
it), mils Voltage, kV VA == Vlt, kV/mil
Equations (9) and (10), derived from analytical 4.7 11.7 2.49
solutions, are identical in form to Equation (6), that was
derived from a computer solution of Laplace's field 6.0 13.0 2.17
equation for the geometry in Figure 41. What is differ-
15.7 17.6 1.12
ent for Equations (9) and (10) are the integer multiplier
of R, and the value of the exponent, both of which are
being dictated by electrode geometries and pairings.
Pairings are the use of two equivalent electrodes as in Using the VA and t data given in Table 24,
Equation (9), or two non-equivalent electrodes as in Equation 11 was solved for K, a, and n by a least-
Equations (6) and (1 0). It is convenient to refer to these squares technique to yield the following:
electrode pairings as "symmetric" or "asymmetric."
VA == 19,173 (t + 3.74)-0.96 (13)
In general, for small values of t or large values of
R, Equations (9) and (10) are of the general form
and, therefore, for t == 0
VA == K (t + a)-n (11 ) S == Emax == K(a)-n == 5404 Vlmil (14)
where a and n are dictated by electrode geometries
and pairings, and when t == 0, In light of the concept being described, it is
tempting to define the value of S == 5404 VImil as the
(12) AC intrinsic dielectric strength of EVA, and to state
whenever this electrical field intensity is reached on an
The experimental observation that VA decreases electrode surface in contact with EVA, the EVA material
with increasing t has been interpreted as a material will experience voltage breakdown.
42
Figure 42 is a log-log plot of VA versus the thick- Note the striking similarity of Equation 15 for
ness term (t + 3.74). This is similar to the historically Mason's polyethylene data measured with asymmetric
empirical data-eorrelation technique of plotting VA versus electrodes and Equation 10, the convergence solution
thickness t on log-log paper, except that here the term a for the asymmetric tip-to-ground electrode configuration.
is included along with t in the abscissa. Again, VA Not only are the values of the exponent n essentially the
decreases with increasing values of t, not because of same, but also, for Mason's data, the value of a, which
any material characteristic, but because of the behavior is equal to R in Equation 10, is the same value of R
of the electrical field distribution associated with derived from the EVA data using symmetric electrodes.
increasing the gap thickness between electrodes It is tempting to define the value of S = 7378 V/mil as
(which happens here to be filled with EVA). the intrinsic dielectric strength of polyethylene.
S = Emax = 7378 Vlmil (16) The effect of aging on the DC intrinsic dielectric
strength is being monitored on EVA specimens being
His data, plotted as VA versus the term (t + 1.20), are aged outdoors on outdoor heating racks at 70, 90, and
shown in Figure 43. 105°C, and in UV-accelerated RS/4 chambers at 50°C
43
10.0 r - - - - - , - - - - - r - - , - . , - . . . , . - - - - , - - - r - - r -....- . - - - - - , - - - . , - - . , - - , - - .
~
I
I-
a
z
UJ
a:
I-
(f)
u 1.0
0:
I-
U
UJ
-'
UJ
0
UJ
a
«
a:
UJ EXPERIMENTAL DATA REPRODUCED
~ FROM REFERENCE 6
1 10 100 1000
THICKNESS TERM (t + 1,20), mils
10,000
8000
6000
GENERAL RELATIONSHIP
'E
3:
>
4:
4000
~, VA = J5.. = K(tr O,50
vt
~', ,
THESE DATA:
I
f- ..........
(9 2000 VA = K(t),O.55
zw DATA POINTS FROM 3M ..........
ex: TECHNICAL LITERATURE ' ..........
f- FOR X-22416 AND X-22417 ..........
en
u 1000 UV-SCREENING ACRYLIC ' ..........
a:f- 800
FILMS ..........
-.....
U ..........
w
POINTS~""
..J 600
W
is
w 400
(9
<t:: DATA
ex: DU PONT TECHNICAL
w
>
<t::
LITERATURE FOR
200 LUCITE
100
1 2 4 6 8 10 2 4 6 8 100 2 4 6 1000
THICKNESS t, mils
VA = (VOLTAGE AT BREAKDOWNl!(THICKNESSl = V/t
44
and at 85°C. There are two RS/4 chambers being The trend indicates a reduction in the DC
operated at 50°C; one with, and one without a periodic dielectric strength with dry aging, apparently being
water spray cycle. These are referred to as RS/4-dry, accelerated by increasing temperature. However, the
and RS/4-wet. The unit operating at 85°C is dry. data are too preliminary to judge the long-term aging
behavior. Interestingly, wet RS/4 aging at 50°C results in
Experimental DC dielectric-strength data of A-9918 an increase in the DC dielectric strength. One hypoth-
EVA from RS/4 aging after 2000 h is shown below: esis is that dry aging may result in an increase in ionic
species that could reduce the dielectric strength,
whereas the water spray cycle results in extraction of
DC Dielectric these ionic species. Thus, the DC dielectric strength
Sample Strength, V/mil increases compared to the control sample. The electrical
insulation aging study continues beyond the preparation
Control 3504 of this document.
45
SECTION VI
2 This remaining sample was too small for mechanical property testing.
47
Table 25. Properties of Cured A-9918 EVA as a Function of Exposure Time to RS/4 UV at 55°C
Total Integrated
Limit Trans- Tensile Strength Elongation at
Lot 2 Specimens mission,a % at Break, Ib/in. 2 Break, %
Control (unaged)
Optical transmission, % 91
Gel content, % 91
After 10 months of thermal aging at 90°C, there was aging at 130°C was yellowing and very intense after
essentially no change in the A-9918 EVA. At 130°C, how- 2 months.
ever, the A-9918 EVA underwent considerable deteriora-
tion from thermal oxidation. The material turned brownl With respect to rooftop applications and with the
orange in color and experienced significant deterioration module design philosophy of shielding the A-9918 EVA
of both optical and mechanical properties. A trace of from harmful UV, the thermal aging behavior at 90°C is
visible yellowing was noted after 1 week of thermal initially encouraging. The 10 month (7200 h) of thermal
48
aging at 90°C corresponds roughly to almost 4 years of of residual Lupersol 101 peroxide curing not
rooftop service at the higher operating temperatures. consumed during the EVA cure. The second yellowing
behavior is associated with thermal andlor UV photooxi-
A two-cell mini-module fabricated with glass, EVA, dation that can eventually become visible to the human
and white Tedlar has been aged up to 18,000 h under eye as was observed in the Springborn aging test.
RS/4 lamps at 85°C. The EVA was a modified A-9918,
having the chemically attachable UV-2098 UV- In general, the EVA aging trends observed at
screening agent in substitution of the Cyasorb UV-531 Springborn and JPL were similar. At JPL, exposure of
(see Table 7). This mini-module included a small cop- Elvax 150 to 6 suns of UV at 30°C resulted in visible
per grid encapsulated within the EVA along with two deterioration within 600 h. The material turned 'yellow
electrically active solar cells connected in series. and developed a sticky surface. Chemical analysis
revealed that some crosslinking had occurred, gener-
Up to 16,000 h at 85°C, there were no visually ating an insoluble gel phase. Specimens of cured
observable changes, and the cells were still producing A-9918 EVA, exposed to 6 suns UV for 1400 h at
the same power as initially measured. At the 18,000 h 30°C, experienced only two detected changes:
inspection point, however, there were indications of cop- depletion of the residual Lupersol 101 peroxide curing
per discoloration and pale yellowing of the EVA. Assum- agent, and trace formation of hydroxyl groups.
ing 1300 h of exposure equates to 1 year outdoors, then
18,000 h calculates out to nearly 17 years outdoors at Samples of cured A-9918 EVA exposed to 6 suns of
85°C. This encouraging aging test is continuing. UV for 500 h at 70°C, and for 800 h at 85°C, survived
in excellent condition. These results indicate that the
2. Jet Propulsion Laboratory Testing Cyasorb UV-531 in the concentration used (0.3 wt %) is
adequate to protect the EVA against UV-activated
JPL has carried out UV aging and thermal aging reactions, even at 6 suns of UV intensity.
(in the dark) of both Elvax 150 and cured A-9918 EVA.
The UV light source of a CER is a 550-W, Conradi Control samples of cured A-9918 EVA, thermally
Hanovia medium-pressure mercury arc lamp surrounded aged in the dark for 400 h at 70°C, and for 800 h at
by a Pyrex water jacket for cooling and for removal of 85°C, were in excellent condition with no detection of
IR and non-terrestrial UV wavelengths. any yellowing associated with thermal oxidation.
Generally, sample temperature control in photo-
Aging of cured A-9918 EVA up to 800 h at 105°C,
thermal aging is difficult because absorbed radiation however, resulted in measurable and observable
causes a rise in temperature, and the higher the acceler-
changes from which fundamental chemical information
ation level, the greater is this uncontrollable temperature
on A-9918 EVA aging could be derived. During the time
rise. CER uses an electrical heater and fan system in period of this test exposure, no yellowing from either
conjunction with appropriate optical filtering to adjust
photooxidation or thermal oxidation was detected for
and control sample temperature precisely (i.e., ± 1°C at the cured A-9918 EVA specimens sandwiched between
50°C, and ±3°C at 135°C). the Pyrex glass covers that were exposed to 6 suns
of UV. Also, chemical analysis of these aged
This device allows UV acceleration of up to 30 suns specimens identified the presence of hydroxyl groups,
while maintaining temperature of the absorbing sur- and barely detected minute traces of acetic acid.
face at 50 to 135°C. The majority of the testings were
carried out at 6 suns of UV intensity at the sample
locations. Assuming a i-sun UV day as 5 h for each However, yellowing of the A-9918 EVA specimens
24 h, then about 300 h of exposure to this UV source thermally aged in the dark at 105°C began early in the
equates with 1 year of outdoor UV exposure. aging test and progressed to become increasingly
intense with continued aging, in contrast to the
The UV-vis.ible IR-absorption spectra of almost all of absence of yellowing from the same specimens
the test specimens were monitored primarily to detect exposed to 6 suns of UV. The interpretation of these
changes in absorption spectra for chemical information, data suggest that UV wavelengths, somewhere
rather than to determine optical transmission relative between 310 and 370 nm, act in turn to photooxidize
to solar-cell performance. Special emphasis was given the yellow thermal oxidation products. The result of
to the sensitive detection of absorbance at 360 and this reaction is yet another degradation product that
400 nm. Absorbance at 360 nm was used to monitor does not have a visible color; in other words, UV
the concentration of Cyasorb UV-531 in the speci- bleaching.
mens, and absorbance at 400 nm was used to have a
more sensitive monitor of material yellowing that may 3. EVA Aging Summary
not be readily visible to the human eye.
Elvax 150 can be degraded by UV photo-
Such monitoring at 400 nm resulted in the detection oxidation, thermal oxidation, and by purely thermal
of what tentatively seems to be two distinctly different decomposition of the acetate groups to acetic acid. As
yellowing processes. The first is a transient yellowing, protection against each of these degradation modes is
generally at low levels of intensity not detectable by the provided, the life and associated peak service
human eye, which is associated with the decomposition temperature of EVA encapsulant can be extended.
49
Analysis of Elvax 150 suggests that the UV wave- 70, 90, and 110°C, and the devices tum on at sunrise
lengths, deleterious to this material and necessary for and off at sunset. This approach eliminates the diffi-
UV photooxidation, are those shorter than 360 nm. culties associated with the irregular spectrum of artificial
Isolation of Elvax 150 from these UV wavelengths, light sources, exposes the specimens to other environ-
with UV-filtering outer covers and/or compounding mental conditions such as rain and pollution, and addi-
additives such as Cyasorb UV-531 , stops UV photo- tionally incorporates a dark cycle. The only acceleration,
oxidation and reduces the aging characteristics of therefore, is in the temperature, all other environmental
Elvax 150 to thermal effects. This basic and very conditions being present in their natural occurrence and
simple concept was established as a fundamental intensity. In summary, the OPTAR device is considered
module design philosophy, and no problem with this to have the following advantages:
concept has been identified in the experimental aging
results to date. (1) Uses natural sunlight, therefore, avoids the
spectral distribution problems encountered
For module applications having daytime peaking with artificial light sources.
temperatures near 55°C, it seems that the life of the
EVA encapsulant is related more to the life of the UV (2) Uses temperature to accelerate the photo-
protection schemes and less to either the thermal thermal reactions and is easily controllable.
behavior of the EVA or thermal protection schemes
(for example, antioxidants). (3) Includes dark-cycle reactions that are a
natural part of field exposure.
The potential for long service life of EVA in modules
at rooftop temperatures (e.g., 85°C) looks encouraging, (4) Includes dew and rain water extraction
but predictions of lifetime would be premature. As at effects.
55°C, UV protection and permanence of the UV protec-
tion is a must. After that, it is not clearly established (5) More closely resembles the environmental
which of the thermally driven processes is most critical. conditions experienced by solar modules.
These processes include the basic thermal oxidation
properties of the Elvax 150, of antioxidants and the (6) Easily accommodates both discrete materials
associated temperature dependency of their protective and entire modules.
induction periods, and the temperature dependence of
any physical loss and depletion of the protective com- (7) May be set at any temperature desired for
pounding additives themselves, such as the UV and the purpose of varying the acceleration rate
thermal stabilization additives. or extrapolating to lower temperatures.
The phenomenon of the UV bleaching of the yellow The initial experiment on these devices using
thermal oxidation products, observed in accelerated polypropylene as a model polymer was impressively
testing at 105°C, suggest possibilities for reducing the successful. Figure 45 depicts an outdoor aging
amount of UV filtering as currently employed to permit characteristic of unstabilized polypropylene that is a
some of the UV bleaching wavelengths to penetrate plot of the materials elongation-at-break (EB) versus
throughout the EVA. In this way, allowable service aging time outdoors. The aging is characterized by
temperatures of the EVA might be slightly raised, permit- two stages: an initial induction period during which EB
ting physical and mechanical deterioration by thermal is virtually unchanged, followed by a second stage
oxidation to proceed at known rates related to lifetime that is an almost precipitous drop in EB. This second
expectations while maintaining optical clarity by the UV
bleaching effect.
50
stage reflects brittle failure of the aged polypropylene At this time, the experiment has been ongoing for
sample, and the time associated with the onset of this 7000 h at 70, 90, and 105°C. 3 The free-standing EVA
precipitous drop is the time-to-brittleness, which is a specimens at 90 and 105°C degraded within 2000 h,
function of temperature. becoming yellow, tacky, and viscous (loss of
mechanical properties).
Tensile bars of unstabilized compression-molded
polypropylene were placed on the OPTAR devices at The mechanical properties of samples removed
the three temperatures and aged until brittle. Figure 46 from the OPTAR devices at a scheduled 3000 h test
is an Arrhenius plot of the time-to-brittleness at each point were unchanged from unaged controls, including
of the OPTAR temperatures, and extrapolation of the the EVA specimens positioned behind FEP. At 4000 h,
data line to lower air temperatures predicts the known the composite samples are observed to be visually and
outdoor aging time of the polypropylene at ambient physically unchanged, including the EVA materials
conditions. behind the non-U V screening FEP film.
2. Module Exposure
IENF IELD, CONNl
Four types of "mini" modules were prepared
EXTRAPOLATION OF
HIGH TEMPERATURE (5 x 8.25 in.), each containing two interconnected
DATA TO AMBIENT 90-mm-diameter solar cells (Solar Power Corp.,
VI YIELDS TYPICAL Woburn, Massachusetts). Each module consisted of
0::
::> LIFE-TIMES OF the following components: (1) glass primed with 11861
~ ABOUT 1/3 TO 1/2-YEAR primer, (2) pottant, (3) the cell pair, and (4) a back-
VI" cover film of white Tedlar (200BS30WH) coated with
VI
w 68040 adhesive. The modules were prepared by a
§ llXXl vacuum bag lamination technique with four different
1::
0::
pottants to test a formulation variable. The four pottants
co are listed in Table 27.
0
~
Figure 46. Natural Outdoor Aging of Polypropylene When aging effects were observed, the first
on the Outdoor Heating Racks at 70, 90, change to be noticed was a slight discoloration in the
and 110°C vicinity of the copper mesh. This was also anticipated
to be the first source of degradation. At 70°C, it is
barely noticeable after 5000 h and is found in two
1. EVA Testing
modules (EVA A9918 and EVA 14747). Apart from this
Following this impressive initial experiment, slight effect, there are no other changes that can be
cured specimens of both A-9918 EVA and 18170 EVA observed in any of the modules at the 70°C/12,000 h
(see Tables 4 and 7 for formulations) were mounted on condition.
the OPTAR devices as free-standing films, and also as
composites behind glass, Tedlar, and Teflon FEP. The At 90°C, the copper discoloration became notice-
Tedlar is the UV-screening film, 100BG30UT, which is able in all modules, but more so for those using
1 mil thick. The Teflon FEP is a non-UV screening film Lupersol-101 as the curing agent, and at 105°C was
which is 1.5 mils thick having one surface that is bond- dramatically visible in all four. In addition to the strong
able (the Du Pont designation is FEP-G). This test with orange/brown color around the copper mesh, signs of
FEP addresses the question of whether or not an EVA flow of the pottant could also be found, especially in the
compounded with weather stabilization additives needs module using EVA A-9918. In the 90°C/12,000 h condi-
an additional UV-screening outer cover. In this case, tion, a few other effects also became noticeable. Some
FEP, a naturally weatherable transparent film (see discoloration (yellowing) of the pottant was found in the
Reference 69), functions only to isolate the EVA from module prepared with EVA A-9918 and EVA 14747.
direct contact with liquid water (rain, dew, etc.), which Both of these formulations were similarly cured with
should eliminate water extraction effects, and also as a Lupersol-101, which suggests that this curing agent
hard, fluorocarbon outer surface for low soil retention. might result in slightly less photostable compositions
3The temperature was lowered to 105°C from 110°C for equipment operating reasons.
51
Table 27. Modules Under Exposure
Pottant
Module Number Formulation Number Description
A = Discolored 1 = No change
B = Broken/fractured 2 = Slight change
C = Corrosion (metallization) 3 = Noticeable
D = Delamination 4 = Moderate
E = Flow/melt 5 = Severe
than the Lupersol-TBEC used in the other two showed meandering cracks because of thermal shock,
compositions. the copper catalyzed degradation of the pottant was
striking, and all the pottants showed some degree of
At the 105°C/12,000-h test point, degradative effects discoloration. These tests are continuing beyond
were now quite noticeable. The glass outer covers all preparation of this document.
52
Eval............,""',
The emphasis and major focus of the FSA Project target with sufficient kinetic energy to achieve some
was the development, demonstration, and transfer to subsurface penetration. This results in strong adhe-
industry of technology applicable to the design and pro- sion of the ion-plated coating in contrast to conven-
duction of large-scale, long-life, and cost-effective PV tional vapor deposition that merely condenses onto a
solar arrays. In the course of the FSA program, a great surface.
variety of candidate materials, processes, and design
concepts were proposed, explored, and evaluated. As Early in the encapsulation task, it was recognized
the FSA Encapsulation Task focused on fewer design that all polymers are permeable to water and, thus,
approaches, it became necessary to drop development elastomeric pottants and plastic films probably would
of a number of alternative materials and process con- not isolate encapsulated solar cells from atmospheric
cepts. These were dropped based on either their imma- moisture. This could be a problem if solar celis were
ture technical status or their lacking in comparative cost found to be sensitive to loss of performance from expo-
effectiveness, considering other available design options. sure to moisture. Therefore, ion plating was investigated
as a possible means of applying a thin and transparent
However, these technologies may be well worth conformal coating as an impervious water barrier over
developing for a different set of PV module design all surfaces of a solar cell. The candidate coating
requirements such as for space, military, or navigation material was aluminum oxide (Figure 47), about 1400A
applications in which ruggedness, portability, or weight thick to also function as an AR coating. Further, it was
factors may be overriding. Therefore, several of these considered that the task of adhesively bonding an elas-
advanced encapsulation concepts are summarized in tomeric pottant to a solar cell would be simplified if the
this section for future reference and potential application. solar cell had only one outer surface chemistry.
100
80
<f. SAPPHIRE
z
o
iii
If)
60
~
If)
z
«
::: 40
l-
I
(:J
::::i
20
ULTRAVIOLET -f---<0>1/ '/1-l.........--INFRAREO
IUVI HRI
LIGHT WAVELENGTH, 1
53
was stopped for two reasons: first, the need for an a UV screening cover for UV-sensitive pottants. The
impervious coating concept vanished when solar cell chemistry of this material permits the incorporation of
manufacturers eliminated the root causes for moisture polymerizable UV screening agents for permanence of
sensitivity and, second, the scale-up and economics the UV screening property. As made, the polymer was
of the ion-plating process for high-volume production dissolved in toluene and the solution then sprayed onto
did not seem attractive at that time. 4 a release paper. After air drying for 30 min at 75°C, the
film could be lifted off the release paper.
One Dow Corning innovation to reduce silicon The Dow Corning film exhibited good resistance to
material use (and cost) involved the machining of dirt and soil retention and was also easily cleaned by
circular recesses into wood substrates to a depth and wiping with a dry or slightly damp cloth. The wiping
diameter slightly larger than the solar cells to be action did not visually' mar or abrade the surface of
encapsulated. Notches were machined between the the film.
circular recesses to accommodate the interconnects.
The recessed and notched wood substrate was then Because of the existence of other commercially
spray coated with the white-pigmented 01-2577 and available UV-screening films, further development of
while still wet (uncured), the interconnected cell string the Dow Corning UV-screening film was suspended.
was positioned in the recesses and notches. An over- but not forgotten.
coat of clear 01-2577 was sprayed on the entire top
surface of the module and cells, and the silicone was
cured for a few hours at 75°C. C. GLASS-REINFORCED CONCRETE
This module design passed the JPL thermal-eyc/e Glass-fiber reinforced concrete panels as potential
test, and one module installed on an outdoor weathering PV module substrates were developed and evaluated
rack at JPL was still functional after 3 years when the by MBAssociates of San Ramon, California. The panels
rack testing was discontinued for other reasons. were 1/4 in. thick and had integral reinforcing ribs on the
back side of the panel. The projected cost was $0.621ft2 ,
01-2577 is commercially available as a solution at a 1982 price quote. The glass-reinforced concrete
a bulk cost of $8.45/lb, which is a 1982 cost quote. panels were intended to be a part of the solar array
The solids content of the solution is 75 % by weight, field structure in addition to serving as a module
which correspond to a dry-solids cost of $11.26/lb, or substrate, thus making them cost effective by fulfilling
5.8¢/ft 2/mil of thickness, assuming no spray loss. two functions.
411linois Tool Works, ENDUREX Division, Elgin, Illinois, LSA Contract 955506.
54
the encapsulant pottant, and a UV-screening film, metal contacts. In this case, process temperatures near
Korad 212, as the top cover. 5 The demonstration the high end of the range are needed and, in some
module was mounted on 6- x 6-in. pressure-treated cases, external pressure must be applied to increase the
wood posts, simulating an array field structure. No amount of deformation.
problems were experienced with these panels. 6
Module designs using ESB have several advantages:
D. ELECTROSTATIC BONDING
(1) The ESB seal is an integral bond between glass
Electrostatic bonding (ESB) is a method of attaching and silicon and, thus, is fully hermetic.
glass sheets to metals or dielectrics without using an
additional adhesive. This technology was developed for (2) There is no pottant between cells and glass
PV application by Spire Corp. It can be used to attach cover to be subject to degradation.
silicon solar cells to a glass superstrate, or to attach one
sheet of glass to another with a dielectric-film interlayer. (3) Cells are attached to a thermal-expansion-
In the bonding process, the glass is heated to a temper- matched glass.
ature high enough to allow ion mobility, but lower than
the softening point of the glass, typically 350 to 650°C. The present process has several requirements:
At this temperature, high voltage is applied across the
glass and the object to be bonded. Rearrangement of (1) Thermal expansion of the glass must match
ions within the glass causes a permanent chemical bond that of the silicon.
to be formed across the interface. The resulting seal is
completely hermetic, and will generally be as strong as (2) Glass deformation for bonding solar cells with
the materials being bonded. raised front metallization require temperatures
Because of the thermal processing involved, the above 500°C. Therefore, the solar cells must
glass used must be a near match in thermal-expansion be able to withstand at least 5 min of exposure
coefficient to the object to be bonded. For silicon to this temperature without significant thermal
solar cells, Pyrex (Corning Type 7740) or Tempax degradation.
(Schott Glass 8330) is acceptable up to a process
temperature of about 400°C, and Corning Type 7070, Figure 48 shows a hybrid module with ESB front
Schott Type 8248, or Owens-Illinois Type ES-1 is lamination and conventional back. This design retains
acceptable up to 650°C. many of the advantages of the fully integral module, at
lower cost: no pottant between glass front and cells,
For adhesion to cells with significant surface metai- expansion-matched superstrate, and hermetic protection
lization, it is necessary to deform the glass around the of cell-front metallization. 7
OUTPUT TERMINAL
INTERCONNECT BACKING} 1-mil ALUMINUM
2-mil MYLAR
~~~~~~~~'~.~/~"~~~~"~/~~~~~~~~~~~~~~~~~~EVA (~HITE)
L -=~-----::Jt"- FRONT
GLASS
SOLAR CELL
ELECTROSTATICALLY
BONDED TO GLASS
Figure 48. Cross-Sectional View of Integral Front, Electrostatically Bonded Module Assembly
55
SECTION VIII
The FSA Encapsulation Task has essentially Project reports referenced, may be further evaluated
achieved its objectives of identifying, developing, for such special applications.
evaluating, and transferring to industry the module
encapsulation technology required for the design and For most residential and utility applications, tempered
manufacture of low-cost, efficient, and durable solar glass sheet has been the structural panel material of
cell modules. choice. Alternate materials that have special charac-
teristics of interest include glass fiber-reinforced concrete
The basic functions of an encapsulation system, panels, silicone/acrylic polymer blends, electrostatic
which are to provide support, optical coupling, electri- bonded modules with no pottant, and wood fiber hard-
cal isolation, and physical protection, were translated board for structural substrate panels.
into required material characteristics. The definition of
these material characteristics and desirable properties B. MATERIAL AGING RESEARCH
led to the assessment of available materials. It also led
to the development of new materials and combinations Most concepts for capturing and converting solar
of materials and processes that met the performance energy involve the use of polymeric materials that will
and economic goals of the FSA Project. be subject to environmental stresses such as heat,
cold, moisture, UV radiation, and physical forces. The
Although the FSA Project was directed toward the development of low-cost materials able to operate in
development of technology for flat-plate silicon solar cell these environments reliably for periods of years was a
modules, the Encapsulation Task has also contributed major challenge of the Encapsulation Task.
significantly to materials technology for other systems
that also need and use environmentally stable polymeric A major scientific contribution of the Encapsulation
elastomers and plastics. Task was the investigation of the mechanisms of long-
term polymer degradation and their effects on material
The Encapsulation Task has succeeded in defining optical, structural, and electrical properties. Through
requirements for each of the encapsulation materials research in understanding the processes and mecha-
for the several different elements of the solar cell nisms of aging degradation, it was possible to select
module encapsulation package required for achieving the most promising candidate material types and pro-
30-year life and high module performance at low cost. vide criteria for modifying such materials to control
degradation and enhance both their performance and
Rapid technical progress, realized during the past fabricability while achieving the FSA Project goals.
several years in solar energy conversion, has estab-
lished the economic viability of PV modules as a Through understanding the degradation mecha-
source of electric power. Many alternate avenues of nisms, it also was possible to develop more valid
progress have been identified and explored in some accelerated testing methods and computer models for
detail. However, much remains to be done in establish- predicting material changes with time and exposure
ing an expanded database of design and manufacturing conditions. These techniques are now available for
experience for optimizing the economics of solar cell more extensive characterization of advanced encap-
power in terms of module cost, reliability, performance, sulation material designs as they are developed.
and safety.
C. POTTANTS
Some of the encapsulation needs and approaches
identified and recommended for further R&D by indus- The testing and chemical modeling studies of the
try and Government are summarized in the following elastomeric pottants EVA, EMA, and PU, as compounded
paragraphs. for solar module use, indicate the potential for 30-year
durability when used behind glass with sound edge
sealing and a plastic film back cover. Defining the
A. ENCAPSULATION SYSTEM DESIGN limits of temperature and the effects of other adverse
ALTERNATIVES environments, the compatibility with various low-cost
cell metallization materials, needs further analysis and
For specialized solar module applications such as field test experience.
military, remote site, limited lifetime, or low-power
uses, other material systems than detailed in this report It has been noted that the hydrocarbon-based poly-
may have advantages of ruggedness, lower weight, mers are more flammable than the fluorocarbon and
resistance to lower or higher temperatures, or lower silicone materials and, therefore, subject to melting and
cost. Therefore, other material systems that have been burning in some of the more severe fire tests applied to
identified and described in this report, or in other FSA roofing materials. Three aspects of this situation
57
needing further study are: (1) the addition of flame technical data to be able to assess the long-term effects
retardants to the pottant, (2) additions to the encap- of combined electrical and environmental stress on the
sulant package to shield the pottant or protect the dielectric strength of polymeric materials. A fundamental
structure under the module from burning, and (3) approach to evaluating and quantifying this characteristic
development of new and appropriate fire safety stan- has been started and needs to be pursued vigorously.
dards and tests.
F. ANTI-SOILING
Each of these aspects of the situation has been
addressed within the FSA Project and the effort should Treatment of the module cover surfaces, whether
be continued. of glass, acrylic, or fluorocarbon, with antisoiling chem-
icals that were defined and evaluated during this effort
has increased module long-term performance by 1 to
D. PRIMERS AND ADHESIVES 6 % , depending on the cover material and the dirtiness
of the environment.
Delamination between the pottant material and the
structural panel and between the pottant and the solar An economical approach is needed to the periodic
cell was one of the first identified solar module cleaning and possible recoating of module covers with
degradaton modes. As a result, primers and surface antisoiling materials. More field testing experience is
preparations were developed and tested that greatly needed to develop a method for minimizing soiling
improved interface bonding within the module for the losses for various solar array locations.
various combinations of materials involved. The long-
term bonding integrity under conditions of high
temperature and moisture exposure seem adequate, G. THIN-FILM PHOTOVOLTAICS
but additional experience is needed to provide a
reliability database. The encapsulation of thin-film PV modules for long-
term, reliable outdoor service represents a future techni-
cal challenge similarly as that carried out for crystalline
E. ELECTRICAL ISOLATION solar cells. This past effort was successfully performed
encompassing an integrated team of Government, univer-
The development of a source of electrical power, sity, and private research facilities focused on clearly
in which the power generating elements (solar cells) defined research and developmental objectives. It is
may be at 1000 V potential and separated by only a believed that this same systematic approach and pro-
few thousandths of an inch from a possible short or gram organization which was effectively used for crystal-
human contact, presents some special material require- line solar cells, as well as the technologies described in
ments. Testing has demonstrated that 5.0 mils of EVA this report, will serve as a solid basis on which to build a
will resist up to 12,000 V DC. However, there is little future thin-film encapsulation program.
58
SECTION IX
References
9. Gupta, A., Photodegradation of Polymeric 19. Bouquet, F.L., Glass for Low-Cost Photovoltaic
Encapsulants of Solar Cell Modules, JPL Solar Arrays, JPL Publication 80-12, JPL Docu-
Document 5101-77, Jet Propulsion Labora- ment 5101-147, DOE/JPL-1012-40, Jet Propul-
tory, Pasadena, California, August 10, 1978. sion Laboratory, Pasadena, California,
February 1, 1980.
10. Gupta, A., Effect of Photodegradation on
Chemical Structure and Surface Charac- 20. Willis, P.B., and Baum, B., Investigations of
teristics of Silicone Pottants Used in Solar Test Methods, Material Properties, and Pro-
59
cesses for Solar-Cell Encapsulants, Annual 30. Phase I Technical Report for Flat-Plate Solar
Report, ERDA/JPL-954527, Springborn Labo- Array Project Contract 955567, Spectrolab,
ratories, Inc., Enfield, Connecticut, July 1980. Inc., Sylmar, California, November 1981.
21. Gupta, A, and Di Stefano, S., "Photocatalytic 31. Liang, R.H., Yavrouian, A., and Gupta, A,
Degradation of a Cross-Linked Ethylene Vinyl "Development of a Weatherable Acrylic
Acetate (EVA) Elastomer," Polymer Preprints, Elastomer for Solar Cell Encapsulation," Pro-
Vol. 21, p. 178, August 1980. ceedings of the 158th Meeting of the
Electrochemical Society, Hollywood, Florida,
22. Cuddihy, E.F., "Theoretical Considerations of Vol. 81-3, pp. 261-265, 1981 .
Soil Retention," Solar Energy Materials, Vol. 3,
pp. 21-33, 1980.. 32. Cuddihy, E.F., Coulbert, C.D., Willis, P.B.,
Baum, B., Garcia, A., and Minning, C.P.,
23. Baum, B., Willis, P.B., and Cuddihy, E.F., "Polymeric Encapsulation Materials for Low-
"Photovoltaic Encapsulation Materials," Pro- Cost Terrestrial Photovoltaic Modules,"
ceedings of the Symposium on Materials and Paper No. 22 in Polymers in Solar Energy
New Processing Technologies for Photo- Utilization, ACS Symposium Series No. 22,
voltaics, Sponsored by the Electrochemical Symposium Series No. 220, March 1982.
Society, Inc., Vol. 81-3, January 1981.
33. Burger, D., and Cuddihy, E.F., "Vacuum
24. Garcia, A., Minning, C.P., and Cuddihy, E.F., Lamination of Photovoltaic Modules," Paper
"A Quantitative Method for Photovoltaic Encap- No. 25 in Polymers in Solar Energy Utilization,
sulation System Optimization," Proceedings of ACS Symposium Series No. 220, March 1982.
the 26th National SAMPE Meeting, April 28-30,
1981, Los Angeles, California. 34. Cuddihy, E.F., Carroll, W.F., Coulbert, C.D.,
Gupta, A., and Liang, R.H., Photo voltaic
25. Garcia, A, Minning, C.P., and Cuddihy, E.F., Module Encapsulation Design and Materials
"An Analytical Approach to Photovoltaic Selection: Volume I, JPL Publication 81-102,
Encapsulation System Design," Proceedings of JPL Document 5101-177, DOE/JPL-1012-56,
the 15th IEEE Photovoltaic Specialists Con- Jet Propulsion Laboratory, Pasadena,
ference, May 1981. California, June 1, 1982.
26. Liang, R.H., Tsay, FD., Moacanin, J., and 35. Liang, R.H., Gupta, A., and Di Stefano, S.,
Gupta, A., "Photodegradation of Poly Photothermal Characterization of Encapsulant
(n-Butylacrylate), A Candidate Encapsulant Materials for Photovoltaic Modules, JPL
Material for Photovoltaic Modules," Proceed- Publication 82-42, JPL Document 5101-210,
ings of the Symposium on Applications of DOE/JPL-1 012-72, Jet Propulsion Labora-
Spectroscopy to Problems in Polymer tory, Pasadena, California, June 1, 1982.
Engineering and Science, 90th National Meet-
ing of AICHE, Houston, Texas, April 1981. 36. Willis, P.B., and Baum, B., Investigation of Test
Methods, Material Properties, and Processes
27. Minning, C.P., Coakley, J.F., Perrygo, C.M., for Solar Cell Encapsulants, Annual Report,
Garcia, A., and Cuddihy, E.F., "Thermal and DOE/JPL-954527-82/83, Springborn Labora-
Optical Performance of Encapsulation Systems tories, Inc., Enfield, Connecticut, July 1982.
for Flat-Plate Photovoltaic Modules," Proceed-
ings of the 15th IEEE Photo voltaic Specialists 37. Borsing, E., Li, S., Nir, Z., Vogl, 0., and
Conference, May 1981. Gupta, A, "Polymerizable [2(2-Hydroxyphenyl)
2H-Benzotriazole] Ultraviolet Stabilizers," Pro-
28. Willis, P.B., and Baum, B., Investigations of ceedings of the IUPAC 28th Macromolecular
Test Methods, Material Properties, and Pro- Symposium, p. 298, July 12-16, 1982,
cesses for Solar-Cell Encapsulants, Annual Amherst, Massachusetts.
Report, Springborn Laboratories, Inc., Enfield,
Connecticut, DOE/JPL 954527-81-15, 38. Duncan, L.B., Minning, C.P., Garcia, A, and
July 1981. CUddihy, E.F., "Empirical Testing of Structural
Deflection Modeling of Large Photovoltaic
29. Cuddihy, E.F., Development of Reduced- Modules," Proceedings of the 16th IEEE
Variable Master Curves for Estimating Tensile Photo voltaic Specialists Conference,
Stresses of Encapsulated Solar Cells Caused September 1982.
by Module Deflection or Thermal Expansion,
JPL Document 51 OH 82, Jet Propulsion Labo- 39. Gupta, A, Liang, R.H., and Vogl, 0., "Photo-
ratory, Pasadena, California, October 1, 1981. oxidative Degradation of Clear Ultraviolet
60
Absorbing Acrylic Copolymer Surfaces," 48. Design and Analysis of Advanced Encapsula-
Physicochemical Aspects of Polymer Surfaces, tion Systems for Terrestrial Photo voltaic
Vol. 1, p. 293, 1982. Modules, 1983 Annual Report, Spectrolab, Inc.,
Sylmar, California.
40. Coulbert, C.D., The Application of Encapsula-
tion Material Stability Data to Photovoltaic 49. Liang, R.H., Chung, S., Clayton, A.,
Module Life Assessment, JPL Publication Di Stefano, S., Oda, K., Hong, S.D., and
83-27, JPL Document 5101-224, DOEI Gupta, A., "Photothermal Degradation of
JPL-1 012-84, Jet Propulsion Laboratory, EthylenelVinylacetate Copolymer," Polymer
Pasadena, California, April 1, 1983. Alloys, Vol. III, pp. 267-278, 1983.
41. Cuddihy, E.F., Coulbert, C.D, Liang, R.H., 50. Liang, R.H., Coulter, D.R., Dao, C., and
Gupta, A., Willis, P.B., and Baum, B., Appli- Gupta, A., "Novel Diagnostic Techniques for
cations of Ethylene Vinyl Acetate as an Early Detection of Photooxidation in
Encapsulation Material for Terrestrial Photo- Polymers," ACS Sym. Series, No. 220,
voltaic Modules, JPL Publication 85-35, JPL Polymers in Solar Energy Utilization,
Document 5101-220, Jet Propulsion Labora- pp. 265-273, 1983.
tory, Pasadena, California, April 15, 1983.
51. Coakely, J.E., Kallis, J.M., Jones, I.R.,
42. Maxwell, H.G, et aI., FSA Field Test Report Cuddihy, E.F., and Garcia, A., "Cell
1980-1982, JPL Publication 83-29, JPL Temperatures in Terrestrial Photovoltaic
Document 5101-215, DOE/JPL-1 012-85, Jet Modules-Effect of Designs Factors," Pro-
Propulsion Laboratory, Pasadena, California, ceedings of the 17th IEEE Photo voltaic
April 15, 1983. Specialists Conference, May 1984.
43. Liang, R.H., Oda, K., Chung, S., Smith, M., and 52. Kallis, J.M., Trucker, D.C., Szeto, R.K.,
Gupta, A., Handbook of Photothermal Test Cuddihy, E.F., and Garcia, A., "Electrical Isola-
Data on Encapsulant Materials, JPL Publication tion Design of Photovoltaic Modules," Pro-
83-32, JPL Document 5101-230, DOEI ceedings of the 17th IEEE Photo vOltaic
JPL-1 012-86, Jet Propulsion Laboratory, Specialists Conference, May 1984.
Pasadena, California, May 1, 1983.
53. Cuddihy, E.F., Photo voltaic Module Encapsula-
44. Cuddihy, E.F., "Surface Soiling: Theoretical tion Design and Materials Selection: Vol. II, JPL
Mechanisms and Evaluation of Low-Soiling Publication 84-34, JPL Document 5101-237,
Coatings," in Proceedings of the Flat-Plate DOE/JPL-1 012-97, Jet Propulsion Laboratory,
Solar Array Project Research Forum on Quan- Pasadena, California, June 1, 1984.
tifying Degradation, E.F. Cuddihy Chairman,
JPL Publication 83-52, JPL Document 5101- 54. Willis, P.B., and Baum, B., Investigations of
231, DOE/JPL-1 012-89, Jet Propulsion Labora- Test Methods, Material Properties, and Pro-
tory, Pasadena, California, June 1, 1983. cesses for Solar-Cell Encapsulants, Annual
Report, DOE/JPL-954527-84-27, Springborn
45. Willis, P.B., and Baum, B., Investigations of Laboratories, Enfield, Connecticut, July 1984.
Test Methods, Material Properties, and Pro-
cesses for Solar-Cell Encapsulants, Annual 55. CUddihy, E.F., and Willis, P.B., Antisoiling
Report, DOE/JPL-954527-83/24, Springborn Technology: Theories of Surface Soiling and
Laboratories, Inc., Enfield, Connecticut, July Performance of Antisoiling Surface Coatings,
1983. JPL Publication 84-72, JPL Document
5101-25, DOE/JPL-1 012-1 02, Jet Propulsion
46. Coulter, DR, Cuddihy, E.F., and Pluedde- Laboratory, Pasadena, California, Novem-
mann, E.P., Chemical Bonding Technology for ber 15, 1984.
Terrestrial Photo voltaic Modules, JPL Publica-
tion 83-86, JPL Document 5101-232, DOEI 56. Kallis, J.M., Trucker, D.C., Cuddihy, E.F., and
JPL-1 012-91, Jet Propulsion Laboratory, Garcia, A., "Method For Calculating Multi-
Pasadena, California, November 15, 1983. dimensional Electrical Fields in Photovoltaic
ModUles," Solar Cells, Vol, 11, pp. 309-330,
47. Cuddihy, E.F., "Breakdown of Organic 1984.
Insulators," in Proceedings of the Flat-Plate
Solar Array Project Research Forum on the 57. Cuddihy, E.F., A Concept for the Intrinsic
Design of Flat-Plate Photo voltaic Arrays for Dielectric Strength of Electrical Insulation
Central Stations, R. Ross, Chairman, JPL Materials, JPL Publication 85-30, JPL Docu-
Publication 84-44, JPL Document 5101-247, ment 5102-252, DOE/JPL-1 012-1 05, Jet
DOE/JPL-1 012-98, Jet Propulsion Laboratory, Propulsion Laboratory, Pasadena, California,
Pasadena, California, December 1983. April 15, 1985.
61
58. Gomez, P.M., Fu, S.K., Gupta, A., and 68. Rainhart, L.G., and Schimmel, Jr., W.P., "Effect
Vogl, 0, "New 2(2-Hydroxyphenyl)2H- of Outdoor Aging on Acrylic Sheet," Solar
Benzotriazole Ultraviolet Absorbers and Their Energy, Vol. 17, p. 259, 1975.
Incorporation into Addition and Condensation
Polymers," Polymer Preprints, Vol. 26, No.1, 69. Weatherability of Teflon Film, Du Pont
pp. 100-101, April 1985. Technical Information Bulletin T-6C, undated.
59. Willis, P.B., Investigation of Test Methods, 70. Rogers, C., "Permeability and Chemical
Material Properties, and Processes for Solar Resistance," Chapter 9 in Engineering Design
Cell Encapsulants, Ninth Annual Report, DOEI for Plastics, E. Baer, Editor, Reinhold
JPL 954527-85-28, Springborn Laboratories, Publishing Co., New York, 1964.
Inc., Enfield, Connecticut, July 1985.
71. Brendlay, W.H., Jr., "Fundamentals of Acrylic
60. Cuddihy, E.F., "Encapsulant Selection and Polymers," Paint and Varnish Production,
Durability Testing Experience," in Reliability Vol. 63, No.7, pp. 19-27, July 1973.
and Engineering of Thin-Film Photo voltaic
Modules, R. Ross and E. Royal, Editors, JPL 72. Boerio, F.J., Gosselin, CA, Dillingham, R.G.,
Publication 85-73, JPL Document 5101-264, and Liu, HW., Analysis of Thin Films on Metal
DOElJPL-1 012-111 , Jet Propulsion Laboratory, Surfaces, J. Adhesion, Vol. 13, pp. 159-176,
Pasadena, California, October 1, 1985. 1981.
61. Garcia, A., Final Technical Report for FSA 73. Hoffman, A.A., and Maag, C.R., "Airborne
Contract 955567, Design, Analysis and Test Particulate Soiling of Terrestrial Photovoltaic
Verification of Advanced Encapsulation Modules and Cover Materials," Proceedings of
Systems, Spectrolab, Inc., Sylmar, California, the Institute of Environmental Sciences, Phila-
DOE/JPL 955567-85-16, November 1985. delphia, Pennsylvania, May 11-14, 1980, Insti-
tute of Environmental Sciences, Mt. Prospect,
62. Koenig, J.L., Boerio, F.J., Plueddemann, E.P., Illinois, 1980.
Miller, J., Willis, P.B., and Cuddihy, E.F.,
Chemical Bonding Technology: Direct 74. Hoffman, A.R., and Maag, C.R., Photo voltaic
Investigation of Interfacial Bonds, JPL Publica- Module Soiling Studies, May 1978 to October
tion 86-6," JPL Document 5101-284, DOEI 1980, JPL Publication 80-87, JPL Document
JPL-1 012-120, Jet Propulsion Laboratory, 5101-131, DOE/JPL-1 012-49, Jet Propulsion
Pasadena, California, January 1986. Laboratory, Pasadena, California, November 1,
1980.
63. Cuddihy, E.F., The Aging Correlation (RH + t):
Relative Humidity (%) + Temperature (OC),
75. Wen, L., An Investigation of the Effect of Wind
JPL Publication 86-7, JPL Document 5101-
Cooling on Photo voltaic Arrays, JPL Publica-
283, DOE/JPL-1 012-121, Jet Propulsion Labo-
tion 82-28, JPL Document 5101-201,
ratory, Pasadena, California, January 15, 1986.
DOElJPL-1 012-69, Jet Propulsion Laboratory,
Pasadena, California, March 1982.
64. Willis, P.B., Investigation of Materials and Pro-
cesses for Solar Cell Encapsulation, Final Raithby, G.D., and Hollands, H.G.T., "A
76.
Report for FSA Contract 954527, DOE/JPL General Method of Obtaining Approximate
954527-86-29, Springborn Laboratories, Inc., Solutions to Laminar and Turbulent Free-
Enfield, Connecticut, August 1986. Convection Problems," Advances in Heat
Transfer, Vol. II, pp. 265-315, ed. T.F. Irvine
and J.P. Hartnett, Academic Press, 1975.
B. LITERATURE CITATIONS
65. Roark, R.J., Formulas for Stress and Strain, 4th 77. McAdams, W.H., Heat Transmission, 3rd
edition, McGraw-Hili, p. 225, 1965. Edition, McGraw-Hili Book Co., Inc., New York,
1954.
66. Stultz, JW., and Wen, L.C., Thermal Perform-
ance Testing and Analysis of Photo voltaic 78. Chen, C.P., Fracture Strength of Silicon Solar
Modules in Natural Sunlight, JPL Document Cells, JPL Document 5101-137, JPL Publica-
5101-31, Jet Propulsion Laboratory, Pasadena, tion 79-102, DOE/JPL-1 012-32, Jet Propulsion
California, July 29, 1977. Laboratory, Pasadena, California, October 15,
1979.
67. Stultz, J.W., Thermal and Other Tests of
Photovoltaic Modules Performed in Natural 79. Conference Records of the IEEE International
Sunlight, JPL Document 5101-76, DOEI Symposia on Electrical Insulation for the Years
JPL-1 012-78/9, Jet Propulsion Laboratory, 1976 to 1982, published and sponsored by the
Pasadena, California, July 31, 1978. IEEE Electrical Insulation Society.
62
80. Tests for Dielectric Breakdown Voltage and 83. Whitehead, S., Dielectric Breakdown of Solids,
Dielectric Strength of Electrical Insulating Oxford at the Clarendon Press, London 1951.
Materials at Commercial Power Frequencies,
ASTM-D-149-64, American Society for Testing 84. Swanson, J.w., and Dall, F.C., "On the Dielec-
Materials, 1970. tric Strength of Synthetic Electrical Insulating
Materials," Conference Record of 1976 IEEE
81. Mason, J.H., "Breakdown of Solid Dielectrics in International Symposium of Electrical Insulation,
Divergent Fields," Proceedings IEEE, Mono- Sponsored by the IEEE Electrical Insulation
graph 127M, 102C, p. 254, Institution of Elec- Society, p. 196, Institute of Electrical and Elec-
trical Engineers, London, 1955. tronics Engineers, Inc., New York, 1976.
82. Ashcraft, A.C., Eichhorn, R.M.,.and
Shaw, R.G., "Laboratory Studies of Treeing 85. Estey, R.S., Measurement of Solar and
in Solid Dielectrics and Voltage Stabilization Simulator Ultraviolet Spectrallrradiance, JPL
of Polyethylene," Conference Record of Document 5101-58, Jet Propulsion Laboratory,
1976 IEEE International Symposium on Elec- Pasadena, California, March 15, 1978.
trical Insulation, Sponsored by the IEEE Elec-
trical Insulation Society, p. 213, Institute of 86. Ware, J.C., "Clear Sky Temperature," pre-
Electrical and Electronics Engineers, Inc., sented at ISES Meeting, Fort Collins, Colorado,
New York, 1976. August 1974.
63
APPENDIX
Glossary
A-I
Prepared by the Jet Propulsion Laboratory, California Institute of Technology,
for the U.S. Department of Energy through an agreement with the National
Aeronautics and Space Administration.
The JPL Flat-Plate Solar Array Project is sponsored by the U.S. Department of
Energy and is part of the National Photovoltaics Program to initiate a major
effort toward the development of cost-competitive solar arrays.
This report was prepared as an account of work sponsored by an agency of the
United States Government. Neither the United States Government nor any
agency thereof, nor any of their employees, makes any warranty, express or
implied, or assumes any legal liability or responsibility for the accuracy, com-
pleteness, or usefulness of any information, apparatus, product, or process
disclosed, or represents that its use would not infringe privately owned rights.
Reference herein to any specific commercial product, process, or service by trade
name, trademark, manufacturer, or otherwise, does not necessarily constitute or
imply its endorsement, recommendation, or favoring by the United States
Government or any agency thereof. The views and opinions of authors expressed
herein do not necessarily state or reflect those of the United States Government
or any agency thereo f.
Dendritic web silicon ribbons are grown to solar-eell The edge-defined film-fed growth silicon ribbons are
thickness. Progress is shown by experimental ribbons grown to solar-eell thickness. A DOE/FSA-sponsored
grown in 1976 and 1978 and a ribbo.1 grown in a research ribbon grown in 1976 is shown next to a
Westinghouse Electric Corporation pilot plant. nine-sided ribbon grown in a Mobil Solar Energy
Corporation funded configuration.
GLASS (STRUCTURAL)
SPACER
POTlANT
POTTANT
Czochralski silicon crystals as grown are Typical superstrate module design is shown with the
sawed into thin circular wafers. (Support for electrically interconnected solar cells embedded in a
this effort was completed in 1981.) laminate that is structurally supported by glass.
Materials and processes suitable for mass production
have been developed using this laminated design.
Later...
House in Carlisle, Massachusetts, with a 7.3-kW A 28-kW array of solar cells for crop irrigation
photo voltaic rooftop array. Excess photo voltaic- during summer, and crop drying during winter
generated power is sold to the utility. Power is (a DOE/University of Nebraska cooperative project).
automatically supplied by the utility as needed.